ii
Contents
Introduction
ix
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
maturity, and
. . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
35
35
36
49
64
67
71
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
79
79
80
81
88
91
95
96
98
12 BlackScholes
12.1 Introduction to the BlackScholes formula . . . .
12.1.1 Call and put option price . . . . . . . . .
12.1.2 When is the BlackScholes formula valid?
12.2 Applying the formula to other assets . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
105
105
105
107
107
iii
.
.
.
.
.
.
1
1
1
11
12
13
19
iv
CONTENTS
12.3
12.4
12.5
12.6
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
107
108
108
110
110
111
111
112
112
112
112
112
113
114
115
115
115
118
119
119
120
121
121
121
125
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
129
129
129
138
140
141
141
142
14 Exotic options: I
14.1 Asian option (i.e. average options)
14.1.1 Characteristics . . . . . . .
14.1.2 Examples . . . . . . . . . .
14.1.3 Geometric average . . . . .
14.1.4 Payo at maturity T . . . .
14.2 Barrier option . . . . . . . . . . . .
14.2.1 Knockin option . . . . . .
14.2.2 Knockout option . . . . . .
14.2.3 Rebate option . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
145
145
145
146
146
147
147
147
147
148
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
CONTENTS
14.2.4 Barrier parity . .
14.2.5 Examples . . . .
14.3 Compound option . . .
14.4 Gap option . . . . . . .
14.4.1 Definition . . . .
14.4.2 Pricing formula .
14.4.3 How to memorize
14.5 Exchange option . . . .
v
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
the pricing formula
. . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
148
148
149
151
151
151
151
152
18 Lognormal distribution
18.1 Normal distribution . . . . . . . . . . . . . . . . . . .
18.2 Lognormal distribution . . . . . . . . . . . . . . . . . .
18.3 Lognormal model of stock prices . . . . . . . . . . . .
18.4 Lognormal probability calculation . . . . . . . . . . . .
18.4.1 Lognormal confidence interval . . . . . . . . . .
18.4.2 Conditional expected prices . . . . . . . . . . .
18.4.3 BlackScholes formula . . . . . . . . . . . . . .
18.5 Estimating the parameters of a lognormal distribution
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
155
155
156
156
157
158
162
162
163
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
165
165
166
167
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
173
173
177
180
184
193
193
197
199
199
199
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
Geometric Brownian
. . . . . . . . . . . . .
205
205
206
207
207
207
209
210
214
vi
CONTENTS
20.2.6 OrnsteinUhlenbeck process . . . . . . . . .
20.3 Definition of the stochastic calculus . . . . . . . . .
20.4 Properties of the stochastic calculus . . . . . . . .
20.5 Itos lemma . . . . . . . . . . . . . . . . . . . . . .
20.5.1 Multiplication rules . . . . . . . . . . . . .
20.5.2 Itos lemma . . . . . . . . . . . . . . . . . .
20.6 Geometric Brownian motion revisited . . . . . . .
20.6.1 Relative importance of drift and noise term
20.6.2 Correlated Ito processes . . . . . . . . . . .
20.7 Sharpe ratio . . . . . . . . . . . . . . . . . . . . . .
20.8 Risk neutral process . . . . . . . . . . . . . . . . .
20.9 Valuing a claim on S a . . . . . . . . . . . . . . . .
20.9.1 Process followed by S a . . . . . . . . . . . .
20.9.2 Formula for S A (t) and E S A (t) . . . . . .
20.9.3 Expected return of a claim on S A (t) . . . .
20.9.4 Specific examples . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
215
216
222
223
223
223
225
225
226
230
233
233
233
234
235
235
21 BlackScholes equation
21.1 Dierential equations and valuation under certainty .
21.1.1 Valuation equation . . . . . . . . . . . . . . . .
21.1.2 Bonds . . . . . . . . . . . . . . . . . . . . . . .
21.1.3 Dividend paying stock . . . . . . . . . . . . . .
21.2 BlackScholes equation . . . . . . . . . . . . . . . . . .
21.2.1 How to derive BlackScholes equation . . . . .
21.2.2 Verifying the formula for a derivative . . . . . .
21.2.3 BlackScholes equation and equilibrium returns
21.3 Riskneutral pricing . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
245
245
245
246
246
246
246
247
250
252
22 Exotic options: II
253
22.1 Allornothing options . . . . . . . . . . . . . . . . . . . . . . . . 253
23 Volatility
24 Interest rate models
24.1 Marketmaking and bond pricing . . . . . . . . . . . . .
24.1.1 Review of duration and convexity . . . . . . . . .
24.1.2 Interest rate is not so simple . . . . . . . . . . .
24.1.3 Impossible bond pricing model . . . . . . . . . .
24.1.4 Equilibrium equation for bonds . . . . . . . . . .
24.1.5 DeltaGamma approximation for bonds . . . . .
24.2 Equilibrium shortrate bond price models . . . . . . . .
24.2.1 Arithmetic Brownian motion (i.e. Merton model)
24.2.2 RendlemanBartter model . . . . . . . . . . . . .
24.2.3 Vasicek model . . . . . . . . . . . . . . . . . . .
24.2.4 CIR model . . . . . . . . . . . . . . . . . . . . .
24.3 Bond options, caps, and the Black model . . . . . . . .
255
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
257
257
257
264
265
268
270
271
271
272
272
274
275
CONTENTS
24.3.1 Black formula . . . . .
24.3.2 Interest rate caplet . .
24.4 Binomial interest rate model
24.5 BlackDermanToy model . .
vii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
275
279
279
283
viii
CONTENTS
Introduction
This study guide is for SOA MFE and CAS Exam 3F. Before you start, make
sure you have the following items:
1. Derivatives Markets, the 2nd edition.
2. Errata of Derivatives Markets. You can download the errata at http://
www.kellogg.northwestern.edu/faculty/mcdonald/htm/typos2e_01.
html.
3. Download the syllabus from the SOA or CAS website.
4. Download the sample MFE problems and solutions from the SOA website.
5. Download the recent SOA MFE and CAS Exam 3 problems.
ix
INTRODUCTION
Chapter 9
Putcall parity
9.1.1
Option on stocks
Notation
t=0
T
S0
ST
K
CEur (K, T )
CEur (K, 0)
PEur (K, T )
PEur (K, 0)
r
F0,T
Putcall parity
The textbook gives the following formula
CEur (K, T ) PEur (K, T ) = P V0,T (F0,T K) = erT (F0,T K)
(9.1)
The textbook explains the intuition behind Equation 9.1. If we set the
forward price F0,T as the common strike price for both the call and the put
1
(9.2)
Later I will explain the intuition behind 9.2. First, lets prove 9.2. The proof
is extremely important.
Suppose at time zero we have two portfolios:
Portfolio #1 consists of a European call option on a stock and P V (K) =
KerT . P V (K) is the present value of the strike price K and r is the
continuously compounded risk free interest rate per year.
Portfolio #2 consists of a European put option on the stock and one share
of the stock with current price S0 .
Both the call and put have the same underlying stock, the same strike price
K, and the same expiration date T . Notice that at time zero Portfolio #1
is worthy CEur (K, T ) + P V (K); Portfolio #2 is worth PEur (K, T ) + S0 .
Since its dicult to compare the value of Portfolio #1 and the value of
Portfolio #2 at time zero, lets compare them at the expiration date T . Well
soon see that the two portfolios have the same value at T .
Payo of Portfolio #1 at the expiration date T
If ST < K If ST K
Call payo is max (0, ST K) 0
ST K
Payo of P V (K)
K
K
Total
K
ST
If you have P V (K) at t = 0, youll have K at T .
Payo of Portfolio #2 at the expiration date T
If ST < K If ST K
Put payo max (0, K ST ) K ST
0
Payo of S0
ST
ST
Total
K
ST
If you have one stock worth S0 at t = 0, youll have one stock at T worth
ST .
You see that Portfolio #1 and Portfolio #2 have identical payos of max (K, ST )
at T . If ST < K, both portfolios are worth the strike price K; if ST K, both
portfolios are worth the stock price ST . Since Portfolio #1 and #2 are worth
the same at T , to avoid arbitrage, they must be worth the same at any time
prior to T . Otherwise, anyone can make free money by buying the lower priced
portfolio and selling the higher priced one. So Portfolio #1 and #2 are worth
the same at time zero. Equation 9.2 holds.
I recommend that from this point now you throw Equation 9.1 away and use
Equation 9.2 instead.
How to memorize Equation 9.2:
Tip 9.1.1. Many candidates have trouble memorizing Equation 9.2. For example, its very easy to write a wrong formula CEur (K, T ) + S0 = PEur (K, T ) +
P V (K). To memorize Equation 9.2, notice that for a call to work, a call must
go hand in hand with the strike price K. When exercising a call option, you
give the call seller both the call certificate and the strike price K. In return, the
call seller gives you one stock. Similarly, for a put to work, a put must go hand
in hand with one stock. When exercising a put, you must give the put seller both
the put certificate and one stock. In return, the put seller gives you the strike
price K.
Tip 9.1.2. Another technique that helps me memorize Equation 9.2 is the
phrase Check (CK) Please (PS). At expiration T , CEur (K, 0)+K = PEur (K, 0)+
ST . Discounting this equation to time zero, we get: CEur (K, T ) + P V (K) =
PEur (K, T ) + S0 .
Example 9.1.1. The price of a 6month 30strike European call option is 12.22.
The stock price is 35. The continuously compounded riskfree interest rate is 8%
per year. Whats the price of a 6month 30strike European put option on the
same stock?
Solution.
K = 30
T = 6/12 = 0.5
S0 = 35
r = 0.08
CEur = 12.22
CEur + P V (K) = PEur + S0
12.22 + 30e0.08(0.5) = PEur + 35,
PEur = 6. 043
Parity if the stock pays discrete dividends
If the stock pays discrete dividend, the parity formula is
CEur (K, T ) + P V (K) = PEur (K, T ) + S0 P V (Div)
(9.3)
This is why we need to subtract the term P V (Div). At expiration, CEur (K, T )+
K = PEur (K, T ) + ST . If we discount ST from T to time zero, well get
S0 P V (Div). If you have one stock worth S0 at time zero, then during [0, T ],
youll receive dividend payments. Then at T , you not only have one share of
stock, you also have the accumulated value of the dividend. To get exactly one
stock at T , you need to have S0 P V (Div) at time zero.
Discounting this equation back to time zero, we get Equation 9.3
Please note that Equation 9.3 assumes that both the timing and the amount
of each dividend are 100% known in advance.
Example 9.1.2. The price of a 9month 95strike European call option is 19.24.
The stock price is 100. The stock pays dividend of $1 in 3 months and $2 in 6
months. The continuously compounded riskfree interest rate is 10% per year.
Whats the price of a 9month 95strike European put option on the same stock?
Solution.
T = 9/12 = 0.75
K = 95
S0 = 100
CEur = 19.24
P V (Div) = e0.1(3/12) + 2e0.1(6/12) = 2. 877 8
r = 0.1
Example 9.1.3. The price of a 9month 83strike European put option is 13.78.
The stock price is 75. The stock pays dividend of $1 in 3 months, $2 in 6 months,
$3 in 9 months, and $4 in 12 months. The continuously compounded riskfree
interest rate is 6% per year. Whats the price of a 9month 95strike European
call option on the same stock?
Solution.
T = 9/12 = 0.75
K = 83
S0 = 75
r = 0.06
PEur = 13.78
P V (Div) = e0.06(3/12) + 2e0.06(6/12) + 3e0.06(9/12) = 5. 794
CEur (K, T ) + P V (K) = PEur (K, T ) + S0 P V (Div)
CEur (K, T ) + 83e0.06(0.75) = 13.78 + 75 5. 794
CEur (K, T ) = 3. 64
Tip 9.1.3. When calculating P V (Div) in Equation 9.3, discard any dividend
paid after the option expiration date. In this example, the $4 is paid in 12
months, which is after the expiration date of the option. This dividend is ignored
when we use Equation 9.3.
(9.4)
Synthetic stock
Rearranging Equation 9.3, we get:
S0 = CEur (K, T ) PEur (K, T ) + P V (K) + P V (Div)
To understand the meaning of Equation 9.5, notice
(9.5)
Meaning at t = 0
buy a Kstrike call expiring in T
sell a Kstrike put expiring in T
buy a zerocoupon bond that pays K at T
buy a zerocoupon bond that pays Div at T
Synthetic Tbill
Rearranging Equation 9.3, we get:
P V (Div) + P V (K) = S0 + PEur (K, T ) CEur (K, T )
(9.6)
According to Equation 9.6, buying one stock, buying a Kstrike put, and
selling a Kstrike call synthetically creates a zero coupon bond with a present
value equal to P V (Div) + P V (K).
Creating synthetic Tbill by buying the stock, buying a put, and selling a
call is called a conversion. If we short the stock, buy a call, and sell a put, we
create a short position on Tbill. This is called a reverse conversion.
Example 9.1.7. Your motherinlaw desperately wants to borrow $1000 from
you for one year. Shes willing to pay you 50% interest rate for using your
money for one year. You really want to take her oer and earn 50% interest.
However, state antiusury laws prohibits any lender from charging an interest
rate equal to or greater than 50%. Since you happen to know the putcall parity,
you decide to synthetically create a loan and circumvent the state antiusury
law. Explain how you can synthetically create a loan and earn 50% interest.
You want to lend $1000 at time zero and receive $1000 (1.5) = 1500 at T = 1.
To achieve this, at time zero you can
have your motherinlaw sell you an asset thats worth $1000
have your motherinlaw sell you a 1500strike, 1year to expiration put
option on the asset
sell your motherinlaw a 1500strike, 1year to expiration call option on
the asset.
Lets see whats happens at T = 1.
If ST 1500, you exercise the put and sell the asset to your motherinlaw
for 1500
if ST 1500, your motherinlaw exercises the call and buys the asset
from you for 1500.
The net eect is that you give your motherinlaw $1000 at time zero and
receive $1500 from her at T = 1.
In this example, you have a long position on the synthetically created Tbill. This is an example of conversion. In contrast, your motherinlaw has a
short position in the synthetically created Tbill. This is an example of reverse
conversion.
10
at t = 0
buy a 6month 45strike European call
short sell one share receiving 50 and invest in a savings account
buy Bond #1 that pays $1 in 3 months
sell Bond #2 that pays $45 in 6 months
11
(4) Bond #1 pays you $1 dividend at the end of Month 3. After receiving
this dividend, you Immediately pay this dividend to the original owner of
the stock you sold short
(3)+(4) will close out your short position on the stock.
The net eect is that if ST 45 you receive K = 45 and you spent the
proceeds from the short sale. This is "giving up an asset (proceeds from the
short sale) and getting the strike price K."
9.1.2
Options on currencies
(9.7)
In Equation 9.7, the underlying asset is 1 euro. The call holder has the
right, at T , to buy the underlying (i.e. 1 euro) by paying a fixed dollar amount
K. The premium of this call option is CEur (K, T ) dollars. Similarly, the put
holder has the right, at T , to sell the underlying (i.e. 1 euro) for a fixed dollar
amount K. The premium of this put option is PEur (K, T ) dollars. x0 is the
price, in dollars, of buying the underlying (i.e. 1 euro) at time zero. r is the
continuously compounded euro interest rate per year earned by the underlying
(i.e. 1 euro).
To understand the term x0 er in 9.7, notice that to have 1 at T , you need
to have er at time zero. Since the cost of buying 1 at time zero is x0
dollars, the cost of buying er at time zero is x0 er dollars.
Equation 9.7 is very similar to Equation 9.4. If you set S0 = x0 and = r ,
Equation 9.4 becomes 9.7. This shouldnt surprise us. Both S0 and x0 refer to
the price of an underlying asset at time zero. S0 is the dollar price of a stock;
x0 is the dollar price of 1 euro. Both and r measure the continuous rate of
reward earned by an underlying asset.
Tip 9.1.4. How to memorize Equation 9.7. Just memorize Equation 9.4. Next,
set S0 = x0 and = r .
Tip 9.1.5. When applying Equation 9.7, remember that K, CEur (K, T ), PEur (K, T ),
and x0 are in U.S. dollars. To remember this, assume that you are living in the
U.S. (i.e. US dollar is your home currency); that your goal is to either buy or
sell the underlying asset (i.e.1) with a fixed dollar amount. Also remember
that r is the euro interest rate on euro money.
Example 9.1.11. The current exchange rate is 1= 1.33 US dollars. The
dollardenominated 6month to expiration $1.2strike European call option on
one euro has a premium $0.1736. The continuously compounded riskfree interest rate on dollars is 6% per year. The continuously compounded riskfree
12
interest rate on euros is 4% per year. Calculate the premium for the dollardenominated 6month to expiration $1.2strike European put option on one euro.
Solution.
CEur (K, T ) + P V (K) = PEur (K, T ) + S0 eT
K = 1.2
T = 0.5
CEur (K, T ) = $0.1736
r = 0.06
= 0.04
S0 = 1.33
0.1736 + 1.2e0.06(0.5) = PEur (K, T ) + 1.33e0.04(0.5)
PEur (K, T ) = 0.03 447
Example 9.1.12. The current exchange rate is $1 = 0.78. The dollardenominated
9month to expiration $1strike European put option on one euro has a premium
$0.0733. The continuously compounded riskfree interest rate on dollars is 7%
per year. The continuously compounded riskfree interest rate on euros is 3% per
year. Calculate the premium for the dollardenominated 9month to expiration
$1strike European call option on one euro.
CEur (K, T ) + P V (K) = PEur (K, T ) + S0 eT
K=1
T = 0.5
PEur (K, T ) = 0.0733
1
r = 0.07
= 0.03
S0 =
= 1. 282 05
0.78
1 0.03(0.75)
e
CEur (K, T )+e0.07(0.75) = 0.0733+
0.78
9.1.3
Options on bonds
This is an option contract where the underlying asset is a bond. Other than
having dierent underlying assets, a stock option and a bond option have no
major dierences. You buy a call option on stock if you think the stock price
will go up; you buy a put option if you think the stock price will drop. Similarly,
you buy a call option on a bond if you think that the market interest rate will
go down (hence the price of a bond will go up); you buy a put option on a bond
if you think that the market interest rate will go up (hence the price of the bond
will go down).
The coupon payments in a bond are like discrete dividends in a stock. Using
Equation 9.3, we get the
CEur (K, T ) + P V (K) = PEur (K, T ) + B0 P V (Coupon)
(9.8)
Please note that in Equation 9.8, in the term P V (Coupon), "coupon" refers
to the coupon payments during [0, T ]. In other words, only the coupons made
during the life of the options are used in Equation 9.8. Coupons made after T
are ignored.
13
Example 9.1.13. A 10year $1,000 par bond pays 8% annual coupons. The
yield of the bond is equal to the continuously compounded riskfree rate of 6%
per year. A 15month $1,000strike European call option on the bond has a
premium $180. Calculate the premium for a 15month $1,000strike European
put option on the bond.
Solution.
The annual eective interest rate is i = e0.06 1 = 6. 184%
1 1.0618410
B0 = 1000 (0.08) a10 +1000v 10 = 1000 (0.08)
+1000 1.0618410 =
0.06184
1132. 50
We need to be careful about calculating P V (Coupon). The bond matures
in 10 years. There are 10 annual coupons made at t = 1, 2, ..., 10. However,
since the option expires at T = 15/12 = 1. 25, only the coupon paid at t = 1 is
used in Equation 9.8.
P V (Coupon) = e0.06 (1000) (0.08) = 75. 34
CEur (K, T ) + P V (K) = PEur (K, T ) + B0 P V (Coupon)
180 + 1000e0.06(1.25) = PEur (K, T ) + 1132. 50 75. 34
PEur (K, T ) = 50. 59
9.1.4
14
the Microsoft stock as the underlying asset and the Google stock as the strike
asset, then its a put option. This option gives you the privilege of selling, at
T = 1, one Microsoft stock for the price of one Google stock.
Definition 9.1.1. An option gives the option holder the privilege, at T , of
surrendering an asset AT and receiving an asset BT (we denote the option as
AT BT ). This is a call option if we view B as the underlying asset and A as
the strike asset (the option holder has the privilege of buying BT by paying AT ).
This is a put option if we view A as the underlying asset and B as the strike
asset (the option holder has the privilege of selling AT and receiving BT ).
Please note that the payo of Option AT BT is max (0, BT AT ).
Example 9.1.14. An option gives the option holder the privilege, at T = 0.25
(i.e. 3 months later), of buying 1 with $1.25. Explain why this option can be
viewed (perhaps annoyingly) as either a call or a put.
Solution.
This option is $1.25 1. The option holder has the privilege, at T = 0.25,
of surrendering $1.25 and receiving 1 (i.e. give $1.25 and get 1).
This is a call option if we view 1 as the underlying asset. The option holder
has the privilege of buying 1 by paying $1.25.
This is also a put option if we view $1.25 as the underlying asset. The option
1
holder has the privilege of selling $1.25 for 1 (i.e. selling $1 for
=
1.25
0.8).
Example 9.1.15. An option gives the option holder the privilege, at T = 0.25,
of buying one Microsoft stock for $35. Explain why this option can be viewed
(perhaps annoyingly) as either a call or a put.
This option is $35 1 M icrosof t stock (give $35 and get 1 Microsoft stock).
If we view the Microsoft stock as the underlying asset, this is a call option. The
option holder has the privilege of buying one Microsoft stock by paying $35.
This is also a put option if we view $35 as the underlying. The option holder
has the privilege of selling $35 for the price of one Microsoft stock.
Example 9.1.16. An option gives the option holder the privilege, at T = 0.25,
of selling one Microsoft stock for $35. Explain why this option can be viewed as
either a call or a put.
This option is 1 M icrosof t stock $35 (give 1 Microsoft stock and get $35).
If we view the Microsoft stock as the underlying asset, this is a put option. The
option holder has the privilege of selling one Microsoft stock for $35. This is
also a call option if we view $35 as the underlying. The option holder has the
privilege of buying $35 by paying one Microsoft stock.
15
(9.9)
Payo of Portfolio #2 at T
If AT BT
Option BT AT AT BT
P V (BT )
BT
Total
AT
If AT < BT
0
BT
BT
16
Stock B and receive Stock A" or "give Stock A and receive Stock B" as a call or
put. Equation 9.9 holds no matter you call the option AT BT or BT AT
a call or put.
Example 9.1.18. Stock A currently sells for $55 per share. It pays dividend of
$1.2 at the end of each quarter. Stock B currently sells for $72 per share. It pays
dividend at a continuously compounded rate of 8% per year. The continuously
compounded riskfree interest rate is 6% per year. A European option gives the
option holder the right to surrender one share of Stock A and receive one share
of Stock B at the end of Year 1. This option currently sells for $27.64. Calculate
the premium for another European option that gives the option holder the right
to surrender one Stock B and receive one Stock A at the end of Year 1.
(AT BT )0 + P V (AT ) = (BT AT )0 + P V (BT )
P V (AT ) = A0 1.2a4i = 55 1.2a4i
i is the eective interest rate per quarter.
i = e0.25r 1 = e0.25(0.06) 1 = 1. 511
3%
1 1.01 511 34
P V (AT ) = 55 1.2a4i = 55 1.2
= 50. 38
0.01 511 3
P V (BT ) = B0 eB T = 72e0.08 = 66. 46
27.64 + 50. 38 = (BT AT )0 + 66. 46
(BT AT )0 = 11. 56
Currency options
Example 9.1.19. Lets go through the textbook example. Suppose that a 1year
dollardenominated call option on 1 with the strike price $0.92 is $0.00337.
The current exchange rate is 1 = $0.9. Whats the premium for a 1year
1
eurodenominated put option on $1 with strike price
= 1. 087?
0.92
First, lets walk through the vocabulary. The phrase "dollardenominated
option" means that both the strike price and the option premium are expressed
in U.S. dollars. Similarly, the phrase "eurodenominated option" means that
both the strike price and the option premium are expressed in euros.
Next, lets summarize the information using symbols. "dollardenominated
call option on 1 with the strike price $0.92" is $0.92 1. The premium for
this option is $0.00337. This is represented by ($0.92 1)0 =$0.0337.
1
= 1. 087" can
"Eurodenominated put option on $1 with strike price
0.92
1
1
be represented by $1
. The premium for this option is $1
0.92
0.92 0
Now the solution
be simple.
should
1
1
1
$1
=
$1 0.92
0.92
0.92 0 0.92
0.92
0
1
1
=
($0.92 1)0 =
$0.0337
0.92
0.92
1
1
1
1
=
$0.0337 =
0.0337
= 0.04 07
$1
0.92 0 0.92
0.92
0.9
17
18
1
1
=b or $1
=b
$1
K 0
K 0
A eurodenominated
It then follows:
$1
1
K
x0 =
$1
1
K
$
0
1
( $K 1)$0
K
(9.10)
1
is the eurocost of "give $1, get ." Since
K
0
1
x0 is the dollar cost of "give
the exchange rate is 1 = $x0 , $1
K 0
1
$1, get
." Similarly, ( $K 1)$0 is the dollarcost of "give $K, get 1."
K
1
1
Equation 9.10 essentially says that the dollar cost of "give $1, get " is
of
K
K
the dollar cost "give $K, get 1." This should make intuitive sense.
1 1
Tip 9.1.6. The textbook gives you the complex formula C$ (x0 , K, T ) = x0 KPf
, ,T .
x0 K
Do not memorize this formula or Equation 9.10. Memorizing complex formulas
is often prone to errors. Just translate options into symbols. Then a simple
solution should emerge. See the next example.
In Equation 9.10,
1
$1
K
19
Example 9.1.20. The current exchange rate is 0.9 per dollar. A European
eurodenominated call on 1 dollar with a strike price 0.8 and 6 months to
expiration has a premium 0.0892. Calculate the price of a European dollardenominated put option on 1 euro with a strike price $1.25.
Just translate the options into symbols. Then youll see a solution.
The current exchange rate is 0.9 per dollar. $1 =0.9 or 1 = $
1
0.9
Eurodenominated call on 1 dollar with strike price 0.8 has a premium 0.0892
(0.8 $1)0 =0.0892
Calculate the price of a dollardenominated put on 1 euro with strike price $1.25
(1 $1.25)0 = $?
1
$1 = 1.25 (0.8 $1)0
(1 $1.25)0 = 1.25
1.25
0
1
= 1.25 0.0892 = 1.25 0.08928 $
= $0.124
0.9
9.1.5
(9.11)
(9.12)
Equation 9.11 and Equation 9.12 are not earthshaking observations. You
shouldnt have trouble memorizing them.
Maximum and minimum price of a call
1. The price of a call option is always nonnegative. CAmer (K, T )
CEur (K, T ) 0. Any option (American or European, call or put) is a
privilege with nonnegative payo. The price of a privilege can never be
negative. The worst thing you can do is to throw away the privilege.
2. The price of a call option cant exceed the current stock price.
S0 CAmer (K, T ) CEur (K, T ). The best you can do with a call option
is to own a stock. So a call cant be worth more than the current stock.
20
(9.13)
(9.15)
Tip 9.1.7. You dont need to memorize Equation 9.13, 9.14, or 9.15. Just memorize basic ideas behind these formulas and derive the formulas from scratch.
Maximum and minimum price of a put
1. The price of a put option is always nonnegative. PAmer (K, T )
PEur (K, T ) 0.
2. The price of a European put option cant exceed the present
value of the strike price. PEur (K, T ) P V (K). The best you can
do with a European put option is to get the strike price K at T . So a
European put cant be worth more than the present value of the strike
price.
3. The price of an American put option cant exceed the strike
price. K CAmer (K, T ). The best you can do with an American put
option is to exercise it immediately after time zero and receive the strike
price K. So an American put cant be worth more than the strike price.
4. The price of a European put option must obey the put call
parity. For a nondividend paying stock, the parity is PEur (K, T ) =
CEur (K, T ) + P V (K) S0 P V (K) S0
Combining 1, 2, and 3, we have:
K PAmer (K, T ) PEur (K, T ) 0
For a nondividend paying stock
(9.16)
21
(9.17)
(9.18)
(9.19)
Tip 9.1.8. You dont need to memorize Equation 9.16, 9.17, 9.18, or 9.19. Just
memorize the basic ideas behind these formulas and derive the formulas from
scratch.
Early exercise of American options
Suppose an American option is written at time zero. The option expires in date
T . Todays date is t where 0 t < T . The stock price at the option expiration
date is ST . Todays stock price is St . The continuously compounded riskfree
interest rate is r per year.
Proposition 9.1.1. Its never optimal to exercise an American call early on a
nondividend paying stock.
If at t you exercise an American call option , your payo is St K.
If at t you sell the remaining
call option, youll
get at least CEur (St , K, T )
max [0, St P V (K)] = max 0, St Ker(T t) , the premium for a European
call option written in t and expiring in T .
22
stock during [t, T ] since the stock doesnt pay any dividend. In addition, by
exercising the call option t, you throw away the remaining call option during
[t, T ].
Example 9.1.21. You purchase a 30strike American call option expiring in 6
months on a nondividend paying stock. 2 months later the stock reaches a high
price of $90. You are 100% sure that the stock will drop to $10 in 4 months.
You are attempted to exercise the call option right now (i.e. at t = 2/12 = 1/6
year) and receive 90 30 = 60 profit. The continuously compounded riskfree
interest rate is 6% per year. Explain why its not optimal to exercise the call
early.
Solution. The problem illustrates the pitfall in common thinking "If you I know
for sure that the stock price is going to fall, shouldnt I exercise the call now and
receive profit right away, rather than wait and let my option expire worthless?"
Suppose indeed the stock price will be $10 at the call expiration date T =
6/12 = 0.5. If you exercise the call early at t = 2/12 = 1/6, youll gain
St K = 90 30 = 60, which will accumulate to 60e0.06(4/12) = 61. 212 at
T = 0.5.
Instead of exercising the call early, you can shortsell the stock at t = 2/12 =
1/6. Then youll receive 90, which will accumulate to 90e0.06(4/12) = 91. 818 at
T = 0.5. Then at T = 0.5, you purchase a stock from the market for 10 and
return it to the brokerage firm where you borrow the stock for short sale. Your
profit is 91. 818 10 = 81. 818, which is the greater than 61. 212 by 81. 818 61.
212 = 20. 606.
Proposition 9.1.2. It might be optimal to exercise an American call option
early for a dividend paying stock.
Suppose the stock pays dividend at tD .
Time 0 ... ... tD ... ... T
Pro and con for exercising the call early at tD .
+. If you exercise the call immediately before tD , youll receive dividend
and earn interest during [tD , T ]
. Youll pay the strike price K at tD , losing interest you could have
earned during [tD , T ]
. You throw away the remaining call option during [tD , T ]. Had you
waited, you would have the call option during [tD , T ]
However, if the accumulated value of the dividend is big enough, then it can
optimal to exercise the stock at tD .
Proposition 9.1.3. If its optimal to exercise an American call early, then the
best time to exercise the call is immediately before the dividend payment.
23
24
(9.20)
25
(9.21)
(9.22)
This is why Equation 9.21 holds. Clearly, CAme (K, T ) CEur (K, T ). This
is because an American call option can always be converted to a European call
option.
To understand why CEur (K, T ) + P V (Div) CAme (K, T ), consider an
American call option and an otherwise identical European call option. Both
call options have the same underlying stock, the same strike price K, the same
expiration date T . The European call option is currently selling for CEur (K, T ).
How much more can the American call option sell for?
The only advantage of an American option over an otherwise identical European call option is that the American call option can be exercised early. The
only good reason for exercising an American call early is to get the dividend.
Consequently, the value of an American call option can exceed the value of an
otherwise identical European call option by no more than the present value of
the dividend. So CEur (K, T ) + P V (Div) CAme (K, T ). A rational person
will pay no more than CEur (K, T ) + P V (Div) to buy the American call option.
So Equation 9.21 holds.
Similarly, an American put is worth at least as much as an otherwise identical
European put.
In addition, the value of an American put exceeds the value of an otherwise
identical European put by no more than K P V (K). The only advantage of
an American put over an otherwise identical European put is that the American
put can be exercised early. The only good reason for exercising an American
put early is to receive the strike price K immediately at time zero (as opposed
to receiving K at T in a European put) and earn the interest on K from time
zero to T . The maximum interest that can be earned on K during [0, T ] is
K P V (K) = K KerT . Consequently, PEur (K, T ) + K P V (K)
PAme (K, T ). Equation 9.22 holds.
Next, we are ready to prove Equation 9.20. CAme (K, T ) PAme (K, T )
reaches its minimum value when CEur (K, T ) reaches it minimum value CEur (K, T )
and PAme (K, T ) reaches its maximum value PEur (K, T ) + K P V (K):
CAme (K, T ) PAme (K, T ) CEur (K, T ) [PEur (K, T ) + K P V (K)]
From the putcall parity, we have:
CEur (K, T ) + P V (K) = PEur (K, T ) + S0 P V (Div)
CEur (K, T ) = PEur (K, T ) + S0 P V (Div) P V (K)
CEur (K, T ) [PEur (K, T ) + K P V (K)]
= PEur (K, T ) + S0 P V (Div) P V (K)
26
Similarly, CAme (K, T )PAme (K, T ) reaches it maximum value when CAme (K, T )
reaches its maximum value and PAme (K, T ) reaches it minimum value.
CAme (K, T ) PAme (K, T ) CEur (K, T ) + P V (Div) PEur (K, T )
CEur (K, T )+P V (Div)PEur (K, T ) = [PEur (K, T ) + S0 P V (Div) P V (K)]
+P V (Div) PEur (K, T )
= S0 P V (Div)
Hence S0 P V (K) CAme (K, T ) PAme (K, T ) S0 P V (Div) K.
Equation 9.20 holds.
Tip 9.1.9. If you cant memorize Equation 9.20, just memorize Equation 9.21,
Equation 9.22, and Equation 9.3. Then can derive Equation 9.20 on the spot.
Example 9.1.22. If the interest rate is zero. Is it ever optimal to exercise an
American put on a stock?
Solution. According to Equation 9.22, if the riskfree interest rate is zero, then
K = P V (K) and PEur (K, T ) PAme (K, T ) PEur (K, T ). This gives us
PAme (K, T ) = PEur (K, T ). So its never optimal to exercise an American put
early if the interest rate is zero.
Example 9.1.23. Is it ever optimal to exercise an American call on a nondividend paying stock?
Solution. According to Equation 9.21, if Div = 0, then CEur (K, T ) CAme (K, T )
CEur (K, T ) or CAme (K, T ) = CEur (K, T ). Its never optimal to exercise the
American call option.
Example 9.1.24.
An American call on a nondividend paying stock with
exercise price 20 and maturity in 5 months is worth 1.5. Suppose that the current
stock price is 1.9 and the riskfree continuously compounded interest rate is 10%
per year. Calculate the nonarbitrage boundary price of the American put option
on the stock with strike price 20 and 5 months to maturity.
Solution.
If you can memorize Equation 9.20, then
S0 P V (K) CAme (K, T ) PAme (K, T ) S0 P V (Div) K
19 20e0.1(5/12) 1.5 PAme (K, T ) 19 0 20
0.184 1.5 PAme (K, T ) 1
1 PAme (K, T ) 1.5 0.184
27
(9.23)
(9.24)
28
European options when the strike price grows over time Typically,
a call or put has a fixed strike price K. However, theres nothing to prevent
someone from inventing a European option whose strike price changes over time.
Consider a European option whose strike price grows with the riskfree interest rate. That is, KT = KerT . What can we say about the price of such a
European option?
For a European option with strike price KT = KerT , a longerlived European
option is at least as valuable as an otherwise identical but shorterlived European
option.
(9.26)
(9.27)
This is why Equation 9.26 holds. Suppose at time zero we buy two European
calls on the same stock. The first call expires at T1 and has a strike price KerT1 .
The second call expires at T2 and has a strike price KerT2 , where T1 > T2 .
Lets choose a common time T1 and compare the payos of these two calls
at T1 . The payo of the longerlived call is max 0, ST1 KerT1 .
The payo of the shorterlived call is calculated as follows. First, we calculate
its payo at T2 . Next, we accumulate this payo from
T2 to T1 . rT
The payo of the shorterlived
call
at
T
is
max
0, ST2 Ke 2 . Next, we
2
rT2
accumulate this payo max 0, ST2 Ke
from T2 to T1 .
As well soon see, its much harder for a shortlived call to have a positive
payo at T1 .
The longerlived call will have a positive payo at ST1 KerT1 at T1 if
ST1 > KerT1 ; all else, the payo is zero.
On the other hand, the shorterlived call will have a positive payo ST1
KerT1 at T1 only if the following two conditions are met
ST2 > KerT2
ST1 > KerT1
29
If ST2 KerT2 , the shorterlived call will expire worthless, leading to zero
payo at T2 , which accumulates to zero payo at T1 .
If ST2 > KerT2 , well receive a positive payo ST2 KerT2 at T2 . If we
accumulate ST2 KerT2 from T2 to T1 , ST2 will accumulate to ST1 and KerT2
to KerT2 er(T1 T2 ) = KerT1 , leading to a total amount ST1 KerT1 at T1 . The
total payo amount ST1 KerT1 is positive if ST1 > KerT1 .
In summary, both calls can reach the common positive payo ST1 KerT1 at
T1 . The longerlived call will reach this payo if ST1 > KerT1 . The shorterlived
call will reach this payo if both ST2 > KerT2 and ST1 > KerT1 . Consequently,
the longlived call has a better payo and should be at least as valuable as the
shorterlived call. Hence Equation 9.26 holds.
If you still have trouble understanding why the longerlived call has a richer
payo, you can draw the following payo table:
The accumulated payo of the shorterlived call at T1
If ST1 KerT1
If ST1 > KerT1
rT2
If ST2 Ke
P ayof f = 0
P ayof f = 0
If ST2 > KerT2 P ayof f = ST1 KerT1 0 P ayof f = ST1 KerT1 > 0
The payo of the longerlived call at T1
If ST1 KerT1 If ST1 > KerT1
rT2
If ST2 Ke
P ayof f = 0
P ayof f = ST1 KerT1 > 0
rT2
If ST2 > Ke
P ayof f = 0
P ayof f = ST1 KerT1 > 0
You can see that the longerlived call has a slightly better payo than the
shorterlived payo. To avoid arbitrage, the longerlived call cant sell for less
than the shorterlived call. Hence Equation 9.26 holds. Similarly, you can prove
Equation 9.27.
Dierent strike price
Proposition 9.1.6. A call (European or American) with a low strike price
is at least as valuable as an otherwise identical call with a higher strike price.
However, the excess premium shouldnt exceed the excess strike price.
0 C (K1 , T ) C (K2 , T ) K2 K1 if K1 < K2
(9.28)
Equation 9.28 should make intuitive sense. The lowerstrike call allows the
call holder to buy the underlying asset by the guaranteed lower strike price.
Clearly, the payo of a lowerstrike call can never be less than the payo of
an otherwise identical but higherstrike call. Consequently, 0 C (K1 , T )
C (K2 , T ). And this is why C (K1 , T ) C (K2 , T ) K2 K1 . The only advantage of a K1 strike call over the K2 strike call is that the guaranteed purchase
price of the underlying asset is K2 K1 lower in the K1 strike call; to buy the
asset, the K1 strike call holder can pay K1 yet the K2 strike call holder will pay
30
(9.29)
The only advantage of a K1 strike European call over the K2 strike European
call is that the guaranteed purchase price of the underlying asset is K2 K1
lower in the K1 strike call at T . Consequently, no rational person will pay more
than CEur (K2 , T ) + P V (K2 K1 ) to buy the K1 strike call.
Please note that CEur (K1 , T )CEur (K2 , T ) P V (K2 K1 ) doesnt apply
to American call options because two American options can be exercised at
dierent dates.
Proposition 9.1.8. A put (European or American) with a higher strike price
is at least as valuable as an otherwise identical put with a lower strike price.
However, the excess premium shouldnt exceed the excess strike price.
(9.30)
(9.31)
Please note that PEur (K2 , T ) PEur (K1 , T ) P V (K2 K1 ) wont apply
to two American put options because they can be exercised at two dierent
dates.
Proposition 9.1.10. A diversified option portfolio is at least as valuable as one
undiversified option. For K1 < K2 and 0 < < 1
31
(9.32)
(9.33)
Please note that Equation 9.32 and Equation 9.33 apply to both European
options and American options.
A call portfolio consists of portion of K1 strike call and (1 ) portion
of K2 strike call. The premium of this portfolio, C (K1 ) + (1 ) C (K2 ),
can be no less than the premium of a single call with a strike price K1 +
(1 ) K2 . Similarly, a call portfolio consists of portion of K1 strike put and
(1 ) portion of K2 strike put. The premium of this portfolio, P (K1 ) +
(1 ) P (K2 ), can be no less than the premium of a single call with a strike
price K1 + (1 ) K2 .
Before proving Equation 9.32, lets look at an example.
Example 9.1.25. (Textbook example 9.8 K2 and K3 switched) K1 = 50, K2 =
65, K3 = 0.4 (50) + 0.6 (65) = 59. C (K1 , T ) = 14, C (K2 , T ) = 5. Explain why
C (59) 0.4C (50) + 0.6C (65).
Payo
59strike call (a)
ST < 50
0
50 ST < 59
0
59 ST < 65
ST 59
ST 65
ST 59
0
0
0
ST 50
0
0.4 (ST 50)
ST 50
0
0.4 (ST 50)
ST 50
ST 65
ST 591
(0.4b + 0.6c) a
32
ST K2
ST K3
0
0
0
ST K1
0
(ST K1 )
ST K1
0
(ST K1 )
ST K1
ST K2
ST K3
b + (1 ) c a
(ST K1 ) 0
(1 ) (K2 ST ) > 0
Please note
(ST K1 ) + (1 ) (ST K2 ) = ST [K1 + (1 ) K2 ] = ST K3
In addition,
(ST K1 ) (ST K3 ) = K3 K1 (1 ) ST
= K1 + (1 ) K2 K1 (1 ) ST = (1 ) K2 (1 ) ST
= (1 ) (K2 ST )
If we buy unit of K1 strike call, buy (1 ) unit of K2 call, and sell 1 unit
of K3 = K1 + (1 ) K2 strike call, well have a nonnegative payo at T . To
avoid arbitrage, the initial cost must be nonnegative. Hence b+(1 ) ca >
0.
Anyway, the call portfolio consisting of portion of K1 strike call and 1
portion of K2 strike call is at least as good as the call with the strike price
K3 = K1 + (1 ) K2 . To avoid arbitrage, C [K1 + (1 ) K2 ] C (K1 ) +
(1 ) C (K2 ).
Example 9.1.27. (Textbook example 9.9 K2 and K3 switched) K1 = 50, K2 =
70, K3 = 0.75 (50) + 0.25 (70) = 55. P (K1 , T ) = 4, P (K2 , T ) = 16. Explain
why P (55) 0.75P (50) + 0.25P (70).
Payo
55strike put (a)
ST < 50
55 ST
50 ST
70 ST
55 ST
0
70 ST
0.25 (70 ST )
0
70 ST
0.25 (70 ST )
0
0
0
(0.75b + 0.25c) a
50 ST < 55
55 ST
Please note
0.75 (50 ST ) + 0.25 (70 ST ) = 55 ST
0.25 (70 ST ) (55 ST ) = 0.75 (ST 50)
3 0.4 (S
T
55 ST < 70
0
ST 70
0
33
ST K2
0
K1 ST
K2 ST
K3 ST
0
K2 ST
(1 ) (K2 ST )
0
K2 ST
(K2 ST )
0
0
ST K3
b + (1 ) c a
(ST K1 ) 0
(K2 ST ) > 0
Please note
(K1 ST ) + (1 ) (K2 ST ) = K3 ST
In addition,
(1 ) (K2 ST ) (K3 ST )
= (1 ) (K2 ST ) K1 (1 ) K2 + ST = (ST K1 )
(ST K1 ) (ST K3 ) = K3 K1 (1 ) ST
= K1 + (1 ) K2 K1 (1 ) ST = (1 ) K2 (1 ) ST
= (1 ) (K2 ST )
The payo is always nonnegative. Consequently, the call portfolio consisting
of portion of K1 strike put and 1 portion of K2 strike put is at least as
good as the put with the strike price K3 = K1 +(1 ) K2 . To avoid arbitrage,
P [K1 + (1 ) K2 ] P (K1 ) + (1 ) P (K2 ).
Exercise and moneyness
Proposition 9.1.11. If its optimal to exercise an option, its also optimal to
exercise an otherwise identical option thats more inthemoney.
This is just common sense. The textbook gives an example. Suppose its
optimal to exercise a 50strike American call on a dividend paying stock. The
current stock price is 70. If its optimal to exercise the American a 50strike
American call, then it must also be optimal to exercise an otherwise identical
call but with a strike price 40.
34
Chapter 10
10.1
15
20
Time 0
?
10
T =1
Time 0
0
T =1
Its hard to directly calculate the price of the call option. So lets build
something that behaves like a call, something that has the same payo pattern
as the call. Suppose at time zero we create a portfolio by buying X stocks and
putting Y dollars in a savings account. We want this portfolio to have the exact
payo as the call.
Y e0.1
30X
20X
Time 0
+
10X
T =1
Y
Time 0
15
=
Y e0.1
T =1
35
?
Time 0
0
T =1
36
In the above diagram, 2X stocks at time zero are worth either 30X or 10X
at T = 1. Putting Y dollars in a savings account at time zero will produce Y e0.1
at T = 1.
We want our portfolio to behave like a call. So the payo of our portfolio
should
as the payo of the call. We set up the following equation:
the same 0.1
30X + Y e = 15
X = 0.75
Y = 10 (0.75) e0.1 = 6. 786
10X + Y e0.1 = 0
Y = 6. 786 means that we borrow 6. 786 at t = 0 and pays 6. 786e0.1 = 7.
5 at T = 1
If at t = 0 we buy 0.75 share of a stock and borrow $6. 786, then at T = 1,
this portfolio will have the same payo as the call. To avoid arbitrage, the
portfolio and the call should have the same cost at t = 0.
C = 20X + Y = 20 (0.75) 6. 786 = 8. 214
Example 10.1.2. Find the price of a 12month European put option on a stock
with strike price $15. The stock currently sells for $20. In 12 months, the
stock can either go up to $30 or go down to $10. The continuously compounded
riskfree interest rate per year is 10%.
Suppose at time zero we create a replicating portfolio by buying X stocks
and investing Y dollars in a savings account. We want this portfolio to have the
exact payo as the put.
Y e0.1
30X
20X
Time 0
10X
T =1
30X + Y e0.1 = 0
10X + Y e0.1 = 5
0
=
Time 0
Y e0.1
T =1
X = 0.25
?
Time 0
5
T =1
Y = 6. 786
So the replicating portfolio is at t = 0 by shortselling 0.25 stock and investing 6. 786 in a savings account.
The price of the put at t = 0 is:
P = 20X + Y = 20 (0.25) + 6. 786 = 1. 786
Lets check whether the putcall parity holds.
C + P V (K) = 8. 214 + 15e0.1 = 21. 787
P + S0 = 1. 786 + 20 = 21. 786
Ignoring the rounding dierence, we get: C + P V (K) = P + S0
10.2
37
Sd
Time h
Sd
Time h
Suppose
you
invest Seh in a savings account, then your wealth at h is
h
rh
simply Se
e = Se(r)h .
Se(r)h
Seh
Se(r)h
Time 0 Time h
To avoid arbitrage, the following condition needs to be met:
38
If the condition is not met, well end up in a weird situation where the stock
is always better o than the savings account or the savings account is always
better o than the stock. Then everyone will invest his money in the better
performing asset, instantly bidding up the price of the lowerperforming asset
and forcing the above condition to be met.
To avoid arbitrage, Su , Sd , and r need to satisfy the following condition:
Su > Se(r)h > Sd
(10.1)
(10.2)
Cd
Time h
4Su + Berh = Cu
4Sd + Berh = Cd
Cu Cd
Su Sd
(10.3)
B = erh
Su Cd Sd Cu
Su Sd
39
(10.4)
Su Serh
Se Sd
rh
Cu +
Cd
(10.5)
C = 4S + B = e
Su Sd
Su Sd
Define
Serh Sd
Su Sd
rh
S
u Se
d = q =
Su Sd
u = p =
(10.6)
(10.7)
(10.8)
(10.9)
(10.10)
(10.11)
S=e
rh
Su Serh
Serh Sd
Su +
Sd
Su Sd
Su Sd
= erh ( u Su + d Sd )
(10.12)
40
call payos discounted at the riskfree interest rate; the stock price at t = 0 is
simply the expected present value of the future stock prices discounted at the
riskfree interest rate. Since the discounting rate is riskfree interest rate, u
and d are called risk neutral probabilities.
Please note that u and d are not real probabilities. They are artificially
created probabilities so that Equation 10.11 and 10.12 have simple and intuitive
explanations.
Example 10.2.1. Using the riskneutral probabilities, find the price of a 12month European call option on a stock with strike price $15. The stock currently
sells for $20. In 12 months, the stock can either go up to $30 or go down to
$10. The continuously compounded riskfree interest rate per year is 10%.
Solution.
Stock price tree
Time 0
T
Su = 30
S = 20
T
Cu = max (0, 30 15) = 15
C =?
Sd = 10
Cd = max (0, 10 15) = 0
Serh Sd
20e0.1(1) 10
u =
=
= 0.605
Su Sd
30 10
d = 1 0.605 = 0.395
Cu = 30 15 = 15
Cd = 0
C = erh ( u Cu + d Cd ) = e0.1(1) (0.605 15 + 0.395 0) = 8. 211
Example 10.2.2. Using the riskneutral probabilities, find the price of a 12month European put option on a stock with strike price $15. The stock currently
sells for $20. In 12 months, the stock can either go up to $30 or go down to
$10. The continuously compounded riskfree interest rate per year is 10%.
Stock price tree
Time 0
T
Su = 30
S = 20
Sd = 10
T
Cu = max (0, 15 30) = 0
Cd = max (0, 15 10) = 5
Serh Sd
20e0.1(1) 10
=
= 0.605
Su Sd
30 10
d = 1 0.605 = 0.395
u =
Cu = 0
Cd = 5
C = erh ( u Cu + d Cd ) = e0.1(1) (0.605 0 + 0.395 5) = 1. 787
41
4eh Su + Berh = Cu
4eh Sd + Berh = Cd
Solving the equations, we get:
Cu Cd
Su Sd
(10.13)
Su Cd Sd Cu
Su Sd
(10.14)
4 = eh
B = erh
Notice whether the stock pays dividend or not, at time zero, we always need
Su Cd Sd Cu
Su Cd Sd Cu
in a savings account, which grows into
to have erh
Su Sd
Su Sd
Cu Cd
dollars at t = h. If the stock doesnt pay dividend, at t = 0 we hold
Su Sd
Cu Cd
shares of stock, which is
shares of stock at t = h ; if the stock pays
Su Sd
Cu Cd
dividend at a continuously compounded rate , at t = 0 we hold eh
,
Su Sd
Cu Cd
which grows
shares of stock at t = h.
Su Sd
Cu Cd
Su Cd Sd Cu
So at time h we need to have
units of stocks and
Su Sd
Su Sd
dollars in a savings account (or a bond), regardless of whether the stock pays
dividend or not.
To see why, suppose our replicating portfolio at t = h (not t = 0) consists
of U shares of stocks and V dollars in a savings account. Then regardless of
whether the stock pays dividend or not, we need to have:
U Su + V = Cu
U Sd + V = Cd
V =
Su Cd Sd Cu
Su Sd
42
S
Su Sd
d
u (r)h
Se
S
S
Se(r)h
d
u
rh
=e
Cu +
Cd
Su Sd
Su Sd
C = 4S + B = erh (u Cu + d Cd )
(10.15)
where
u =
Se(r)h Sd
Su Sd
(10.16)
d =
Su Se(r)h
Su Sd
(10.17)
e(r)h d
ud
(10.18)
d =
u e(r)h
ud
(10.19)
Tip 10.2.1. If you dont want to memorize Equation 10.12, 10.15, 10.16, 10.17,
just set up the replication portfolio and calculate 4 and B from scratch.
Example 10.2.3. Find the price of a 12month European call option on a stock
with strike price $15. The stock pays dividends at a continuously compounded
rate 6% per year. The stock currently sells for $20. In 12 months, the stock can
either go up to $30 or go down to $10. The continuously compounded riskfree
interest rate per year is 10%.
Stock price tree
Time 0
T
Su = 30
S = 20
Sd = 10
T
Cu = max (0, 30 15) = 15
Cd = max (0, 10 15) = 0
43
T
(4u , Bu ) = (0.75, 7. 50)
T
Su = 30
S = 20
Sd = 10
T
Cu = max (0, 15 30) = 0
Cd = max (0, 15 10) = 5
44
T
(4u , Bu ) = (0.25, 7. 50)
Solution.
T
Su = 30
S = 20
Time 0
C =?
Sd = 10
Replicating portfolio
Time 0
(4, B) = (0.706 32, 6. 786 28)
45
T
Cu = max (0, 30 15) = 15
Cd = max (0, 10 15) = 0
T
(4u , Bu ) = (0.75, 7. 50)
(4d , Bd ) = (0.75, 7. 50)
There are two calls. One is in the market selling for $8 at t = 0. The other is
a synthetic call, which consists, at t = 0, of holding 0.706 3 stock and borrowing
$6. 786 3 at riskfree interest rate. The synthetic call sells for $7. 34 at t = 0.
These two calls have identical payos at t = 1.
To make a riskless profit, we buy low and sell high. At t = 0, we sell a
call for $8 (sell high). Then at t = 1, if the stock price is $30, the call holder
exercises the call and our payo is 15 30 = 15; if the stock is $10, the call
expires worthless and our payo is zero.
15
0
Time 0 Time t = 1
Payo of a written call
Simultaneously, at t = 0 we buy 0.706 3 stock and borrow 6. 786 3 dollars
at riskfree interest rate (buy low). This costs us 0.706 3 (20) 6. 786 3 = 7.
34 at t = 0. At t = 1, our initial 0.706 3 stock becomes 0.706 3e0.06(1) stock
and our initial debt 6. 786 3 grows into 6. 786 3e0.1(1) . Our portfolio is worth
0.706 3e0.06(1) S1 6. 786 3e0.1(1) .
If S1 = 30, our portfolio is worth
0.706 3e0.06(1) (30) 6. 786 3e0.1(1) = 15
If S1 = 10, our portfolio is worth
0.706 3e0.06(1) (10) 6. 786 3e0.1(1) = 0
15
7.34
Time 0
0
Time t = 1
46
Example 10.2.6. A 12month European call option on a stock has strike price
$15. The stock pays dividends at a continuously compounded rate 6% per year.
The stock currently sells for $20. In 12 months, the stock can either go up to
$30 or go down to $10. The continuously compounded riskfree interest rate per
year is 10%. This call currently sells for $7. Design an arbitrage strategy.
Stock price tree
Time 0
T
Su = 30
S = 20
Time 0
C =?
Sd = 10
Replicating portfolio
Time 0
(4, B) = (0.706 32, 6. 786 28)
T
Cu = max (0, 30 15) = 15
Cd = max (0, 10 15) = 0
T
(4u , Bu ) = (0.75, 7. 50)
(4d , Bd ) = (0.75, 7. 50)
There are two calls. One is in the market selling for $7 at t = 0. The other is
a synthetic call, which consists, at t = 0, of holding 0.706 3 stock and borrowing
6. 786 3 dollars at riskfree interest rate. The synthetic call sells for $7. 34 at
t = 0. These two calls have identical payos at t = 1.
To make a riskless profit, we buy low and sell high. At t = 0, we buy a
call for $7 (buy low). Then at t = 1, if the stock price is $30, the call holder
exercises the call and our payo is 30 15 = 15; if the stock is $10, the call
expires worthless and our payo is zero.
15
7
0
Time 0 Time t = 1
Payo of a purchased call
Simultaneously, at t = 0 we sell the replicating portfolio. We short sell
0.706 3 stock and lend 6. 786 3 dollars at riskfree interest rate (sell high). We
gain 0.706 3 (20) 6. 786 3 = 7. 34 at t = 0.
At t = 1, our initially borrowed 0.706 3 stock becomes 0.706 3e0.06(1) stock
and our lent principal 6. 786 3 grows into 6. 786 3e0.1(1) . Our portfolio is worth
6. 786 3e0.1(1) 0.706 3e0.06(1) S1 .
If S1 = 30, our portfolio is worth
6. 786 3e0.1(1) 0.706 3e0.06(1) (30) = 15
If S1 = 10, our portfolio is worth
47
15
Time 0
0
Time t = 1
u Su + d Sd =
u Su + d Sd = Se(r)h = F0,h
(10.20)
(10.21)
48
According to Equation 10.20, the undiscounted stock price equals the forward price under the risk neutral probability. If a problem gives you a forward
price, you can use Equation 10.20 to calculate the riskneutral probability.
Constructing a binomial tree
Suppose we are standing at time t. If we can be 100% certain about the stock
price at time t + h, then investing in stocks doesnt have any risk. Then stocks
must earn a riskfree interest rate. Since the stock already pays dividends at
rate , to earn a risk free interest rate, the stock price just needs to grow at
the rate of r . Hence the stock price at t + h is Se(r)h , which is just the
forward price Ft,t+h because Ft,t+h is also equal to Se(r)h .
However, the stock price is a random variable and we generally cant be
100% certain about a stocks future price. To incorporate uncertainty, we use
Formula 11.16 in Derivatives Markets:
ln
St+h
= (r ) h h
St
(Textbook 11.16)
St+h
is the continuously compounded rate of return
St
during [t, t+ h]. This return consists of a known element (r ) h and a random
element h, where is the annualized standard deviation of the continuously
compounded stock return. The variance of the stock return in one year is 2 .
The variance during the interval [t, t + h] (which is h year long) is 2 h and this
is why. [t, t + h] can be broken down into h intervals, with each interval being
one year long. Assume stock return during each year is independent identically
distributed. The total return during [t, t + h] is the sum of the returns over h
intervals. Then the total variance is just the sum of the variance over h intervals,
2 h.
In the above equation, ln
e h . In the
binomial
So St+h = St e(r)h h = Se(r)h h = Ft,t+h
(r)h+ h
h
model, the stock price
either goes up to Se
= Ft,t+h e
or goes
uSt = Ft,t+h e
dSt = Ft,t+h e
(10.22)
(10.23)
If we set volatility to zero, then Equation 10.22 and 10.23 becomes uSt =
dSt = Ft,t+h . This means that if the stock price is 100% certain, then the stock
price is just the forward price.
Apply Equation 10.21 to Equation 10.22 and 10.23, we get:
u = e(r)h+
d = e(r)h
10.2.1
49
(10.24)
(10.25)
Example 10.2.7. Lets reproduce Derivatives Markets Figure 10.4. Here is the
recap of the information on a European call option. The current stock price is 41.
The strike price K = 40. The annualized standard deviation of the continuously
compounded stock return is = 30%. The continuously compounded riskfree
rate per year is r = 8%. The continuously compounded dividend rate per year
is = 0%.The option expiration date is T = 2 years. Use a 2period binomial
tree to calculate the option premium.
T
Each period is h =
= 1 year long.
2
Step 1
Draw a stock price tree.
2
Suu = 87.6693
Sud = 48.1139
Sdd = 26.4055
Step 2
Draw the terminal payos at T = 2.
47.6693 = max (0, 87.6693 40)
8.1139 = max (0, 48.1139 40)
Option Payo
Period 0
1
Cuu
2
= 47.6693
Cu
C =?
Cud = 8.1139
Cd
Cdd = 0
Step 3
Work backward from right to left (called backwardization). Calculate the premium using the formula
C = erh (u Cu + d Cd )
e(r)h d
e(0.080)1 0.802 52
u =
=
= 0.425 56
ud
1. 462 28 0.802 52
50
2
Cuu = 47.6693
Cu = 23.0290
C = 10.7369
Cud = 8.1139
Cd = 3.1875
Cdd = 0
51
Step 4
Period 0
2
Suu = 87.6693
Period 0
Sud = 48.1139
C = 10.7369
Su = 59.9537
S = 41
1
Cu = 23.0290
Sd = 32.9033
Cud = 8.1139
Cd = 3.1875
Sdd = 26.4055
Period 0
Cdd = 0
1
Cuu Cud
47.6693 8.1139
= e0(1)
= 1.0
Suu Sdd
87.6693 48.1139
Cud Cdd
8.1139 0
4d = eh
= e0(1)
= 0.373 8
Sud Sdd
48.1139 26.4055
Cu Cd
23.0290 3.1875
4 = eh
= e0(1)
= 0.733 5
Su Sd
59.9537 32.9033
4u = eh
Bu = erh
2
Cuu = 47.6693
52
Step 5
Verify that the portfolio replicates the premium tree. Here is
recap of the information:
Period 0
1
2
0
1
2
Suu = 87.6693
Cuu = 47.6693
Su = 59.9537
Cu = 23.0290
S = 41
Sud = 48.1139 C = 10.7369
Cud = 8.1139
Sd = 32.9033
Cd = 3.1875
Sdd = 26.4055
Cdd = 0
Period 0
1
(4u , Bu ) = (1.0, 36.9247)
(4d , Bd ) = (0.3738, 9.1107)
53
Tip 10.2.3. For a multibinomial tree, using the risk neutral probability to find
the premium is faster than using the replicating portfolio. The risk neutral probabilities u and d are constant cross nodes. However, the replicating portfolio
(4, B) varies by node.
Tip 10.2.4. For European options, you can calculate the premium using the
terminal payos. This is how to quickly find the premium for this problem.
Node at T = 2
uu
ud
dd
Total
Payo at T
Cuu = 47.6693
Cud = 8.1139
Cdd = 0
The premium is the expected present value of the terminal payo using the
risk neutral probability.
1
year long.
3
2 There
54
d = e(r)h h = e(0.080)1/30.3
Stock price
Period 0
1
2
1/3
= 0.863 693
3
Su3 = 74.6781
Su2 = 61.1491
Su2 d = 52.8140
Su = 50.0711
S = 41
Sud = 43.2460
Sd2 u = 37.3513
Sd = 35.4114
2
Sd = 30.5846
Sd3 = 26.4157
Calculate the premium by working backward from right to left.
Period 0
2
Cu2 = 22. 201 6
Cu = 12. 889 5
C = 7. 073 9
Cud = 5. 699 5
Cd = 2. 535 1
Cd2 = 0
rh
3
Cu3 = max (0, 74.6781 40) = 34. 678 1
Cu2 d = max (0, 52.8140 40) = 12. 814
Cd2 u = max (0, 37.3513 40) = 0
Cd3 = max (0, 26.4157 40) = 0
C=e
( u Cu + d Cd )
e(r)h d
e(0.080)1/3 0.863 693
u =
=
= 0.456 806
ud
1. 221 246 0.863 693
d = 1 u = 1 0.456 806 = 0.543 194
55
Now Im going to tell you a calculator shortcut I used when I was preparing
for the old Course 6. The above calculations are intense and prone to errors.
To quickly and accurately calculate the premium at each node, use TI30X IIS
calculator because TI30X IIS allows you to modify formulas easily.
For example, to calculate Cu2 = erh ( u Cu3 + d Cu2 d ), enter
e(0.08/3)(0.456 806 34. 678 1 + 0.543 194 12. 814
Please note for TI30 IIS, the above expression is the same as
e(0.08/3)(0.456 806 34. 678 1 + 0.543 194 12. 814)
In other words, you can omit the ending parenthesis ")". I tell you this
because occasionally people emailed me saying they discovered a typo. This is
not a typo.
Now you have entered the formula
e(0.08/3)(0.456 806 34. 678 1 + 0.543 194 12. 814.
Press "=" and you should get: 22.20164368
Next, to calculate Cud = erh (ud Cu2 d + d Cd2 u ), you dont need to enter
a brand new formula. Just reuse the formula
e(0.08/3) (0.456 806 34. 678 1 + 0.543 194 12. 814)
Change 34. 678 1 to 12. 8140 (so 0.456 806 34. 678 1 becomes 0.456 806 12.
8140).
Change 12. 814 into 00.000 (so 0.543 19412. 814 becomes 0.543 19400.000).
Now the modified formula is
e(0.08/3) (0.456 806 12. 8140 + 0.543 194 00.000)
Press "=" and you should get: 5.6994813
To calculate Cu = erh (u Cu2 + d Cud ), once again reuse a previous formula. Change the formula
e(0.08/3) (0.456 806 12. 8140 + 0.543 194 00.000)
into
e(0.08/3) (0.456 806 22. 201 6 + 0.543 194 05. 699 5)
Press "=" and you should get:12.8894166
Reusing formulas avoids the need to retype erh , u , and d (these three
terms are constant across all nodes) and increases your speed and accuracy. By
reusing formulas, you should quickly find C = 7. 073 9.
56
3
Su3 = 74.6781
Su2 = 61.1491
Su2 d = 52.8140
Su = 50.0711
S = 41
Sud = 43.2460
Sd2 u = 37.3513
Sd = 35.4114
2
Sd = 30.5846
Sd3 = 26.4157
Period 0
3
Cu3 = 34. 678 1
Cd = 2. 535 1
Cd2 u = 0
Cd2 = 0
Cd3 = 0
Period 0
2
(4, B)u2 = (1, 38. 947 4)
(4, B)ud = (0.828 7, 30. 138 6)
(4, B)d2 = (0, 0)
Cu3 Cu2 d
34. 678 1 12. 814
= e0(1/3)
=1
Su3 Su2 d
74.6781 52.8140
Su3 Cu2 d Su2 dCu2 d
74.6781 (12. 814) 52.8140 (34. 678 1)
= e0.08(1/3)
=
Bu2 = erh
Su3 Su2 d
74.6781 52.8140
38. 947 4
4u2 = eh
Cu2 d Cd2 u
12. 814 0
= e0(1/3)
= 0.828 7
2
2
Su d Sd u
52.8140 37.3513
2
2
Su dCd2 u Sd uCu2 d
52.8140 (0) 37.3513 (12. 814)
= e0.08(1/3)
=
Bud = erh
2
2
Su d Sd u
52.8140 37.3513
30. 138 6
4ud = eh
4d2 = eh
Cd2 u Cd3
=0
Su2 d Sd3
57
Cu2 Cud
22. 201 6 5. 699 5
= e0(1/3)
= 0.921 8
Su2 Sud
61.1491 43.2460
Su2 Cud SudCu2
61.1491 (5. 699 5) 43.2460 (22. 201 6)
Bu = erh
= e0.08(1/3)
=
2
Su Sud
61.1491 43.2460
33. 263 6
4u = eh
Cud Cd2
5. 699 5 0
= e0(1/3)
= 0.450 1
Sud Sd2
43.2460 30.5846
SudCd2 Sd2 Cud
43.2460 (0) 30.5846 (5. 699 5)
= e0.08(1/3)
=
Bd = eh
2
Sud Sd
43.2460 30.5846
13. 405 2
4d = eh
Cu Cd
12. 889 5 2. 535 1
= e0(1/3)
= 0.706 3
Su Sd
50.0711 35.4114
Cu Cd
50.0711 (2. 535 1) 35.4114 (12. 889 5)
B = erh
= e0.08(1/3)
=
Su Sd
50.0711 35.4114
21. 885 2
4 = eh
If our goal is to calculate the premium without worrying about the replicating
portfolio, then we just need to know the risk neutral probability and the terminal
payo.
Node
u3
u2 d
ud2
d3
Payo
Cu3 = 34. 678 1
Cu2 d = 12. 814
Cd2 u = 0
Cd3 = 0
2
2
3 u d = 3 0.456 806 0.543 194
3 u 2d = 3 (0.456 806) 0.543 1942
3d = 0.543 1943
The option premium is just the expected present value of the terminal payos
using the riskneutral probability.
= e0.08(1) 34. 678 1 0.456 8063 + 12. 814 3 0.456 8062 0.543 194
= 7. 073 9
58
Solution.
Each period is h =
1
year long.
3
2
Cu2 = 0
Cu = 0.740 9
C = 2. 998 5
Cud = 1. 400 9
Cd = 5. 046 2
Cd2 = 8. 362 9
3
Cu3 = max (0, 40 74.6781) = 0
Cu2 d = max (0, 40 52.8140) = 0
Cd2 u = max (0, 40 37.3513) = 2. 648 7
Cd3 = max (0, 40 26.4157) = 13. 584 3
C = erh ( u Cu + d Cd )
e(r)h d
e(0.080)1/3 0.863 693
u =
=
= 0.456 806
ud
1. 221 246 0.863 693
d = 1 u = 1 0.456 806 = 0.543 194
0
Cud = erh ( u Cu2 d + d Cd2 u ) = e0.08(1/3) (0.456 806 0 + 0.543 194 2. 648 7) =
1. 400 9
Cd2 = erh (u Cd2 u + d Cd3 ) = e0.08(1/3) (0.456 806 2. 648 7 + 0.543 194 13. 584 3) =
8. 362 9
Cu = erh ( u Cu2 + d Cud ) = e0.08(1/3) (0.456 806 0 + 0.543 194 1. 400 9) =
0.740 9
Cd = erh ( u Cud + d Cd2 ) = e0.08(1/3) (0.456 806 1. 400 9 + 0.543 194 8. 362 9) =
5. 046 2
59
2
(4, B)u2 = (0, 0)
Cu3 Cu2 d
00
= e0(1/3)
=0
Su3 Su2 d
74.6781 52.8140
Su3 Cu2 d Su2 dCu2 d
74.6781 (0) 52.8140 (0)
= erh
= e0.08(1/3)
=0
Su3 Su2 d
74.6781 52.8140
4u2 = eh
Bu2
Cu2 d Cd2 u
0 2. 648 7
= e0(1/3)
= 0.171 3
2
2
Su d Sd u
52.8140 37.3513
2
2
Su dCd2 u Sd uCu2 d
52.8140 (2. 648 7) 37.3513 (0)
= erh
= e0.08(1/3)
=
2
2
Su d Sd u
52.8140 37.3513
4ud = eh
Bud
8. 808 8
Cd2 u Cd3
2. 648 7 13. 584 3
= e0(1/3)
= 1.0
Su2 d Sd3
37.3513 26.4157
2
3
Su dCd3 Sd Cd2 u
37.3513 (13. 584 3) 26.4157 (2. 648 7)
= e0.08(1/3)
=
Bd2 = erh
Su2 d Sd3
37.3513 26.4157
38. 947 4
C 2 Cud
0 1. 400 9
= e0(1/3)
= 0.07 82
4u = eh u2
Su Sud
61.1491 43.2460
2
Su Cud SudCu2
61.1491 (1. 400 9) 43.2460 (0)
= e0.08(1/3)
=
Bu = erh
2
Su Sud
61.1491 43.2460
4. 659
4d2 = eh
Cud Cd2
1. 400 9 8. 362 9
= e0(1/3)
= 0.549 9
Sud Sd2
43.2460 30.5846
SudCd2 Sd2 Cud
43.2460 (8. 362 9) 30.5846 (1. 400 9)
= e0.08(1/3)
=
Bd = eh
Sud Sd2
43.2460 30.5846
24. 517 3
4d = eh
Cu Cd
0.740 9 5. 046 2
= e0(1/3)
= 0.293 7
Su Sd
50.0711 35.4114
Cu Cd
50.0711 (5. 046 2) 35.4114 (0.740 9)
= e0.08(1/3)
= 15.
B = erh
Su Sd
50.0711 35.4114
039 5
4 = eh
60
If our goal is to calculate the premium without worrying about the replicating
portfolio, then we just need to know the risk neutral probability and the terminal
payo.
Node
u3
u2 d
ud2
d3
Payo
Cu3 = 0
Cu2 d = 0
Cd2 u = 2. 648 7
Cd3 = 13. 584 3
2
2
3u d = 3 0.456 806 0.543 194
3 u 2d = 3 (0.456 806) 0.543 1942
3d = 0.543 1943
The option premium is just the expected present value of the terminal payos
using the riskneutral probability.
= e0.08(1) 2. 648 7 3 0.456 806 0.543 1942 + 13. 584 3 0.543 1943 =
2. 998 5
1
year long.
3
61
3
Vu3 = max (0, 40 74.6781) = 0
Vu2
Vu
V
Vud
Vd
Vd2
Step 3
Calculate the value of the American put one step left to the
expiration.
An American put can be exercised immediately. The value of an American
option if exercised immediately is called the exercise value (EV ) or intrinsic
value. At Period 2, we compare the value calculated using backwardization and
the exercise value. We take the greater of the two as the value of the American
put.
Cud = erh (u Cu2 d + d Cd2 u ) = e0.08(1/3) (0.456 806 0 + 0.543 194 2. 648 7) =
1. 400 9
Cd2 = erh ( u Cd2 u + d Cd3 ) = e0.08(1/3) (0.456 806 2. 648 7 + 0.543 194 13. 584 3) =
8. 362 9
The exercise values at Period 2 are:
3
Vu3 = 0
Vu2 = 0
Vu
V
Vu2 d = 0
Vud = 1. 400 9
Vd
Vd2 u = 2. 648 7
Vd2 = 9. 415 4
Vd3 = 13. 584 3
62
Step 3
Move one step to the left. Repeat Step 2. Compare the backwardized value and the exercise value. Choose the greater.
The backwardized values are:
Cu = e0.08(1/3) (0.456 806 0 + 0.543 194 1. 400 9) = 0.740 9
Cu = e0.08(1/3) (0.456 806 1. 400 9 + 0.543 194 9. 415 4) = 5. 602 9
The exercise value at Period 1 is:
EVu = max (0, K Su ) = max (0, 40 50.0711) = 0
EVd = max (0, K Sd ) = max (0, 40 35.4114) = 4. 588 6
Vu = max (Cu , EVu ) = max (0.740 9, 0) = 0.740 9
Vud = max (Cd , EVd ) = max (5. 602 9, 4. 588 6) = 5. 602 9
Hence we have:
Period 0
3
Vu3 = 0
Vu2 = 0
Vu = 0.740 9
V
Vu2 d = 0
Vud = 1. 400 9
Vd = 5. 602 9
Vd2 u = 2. 648 7
Vd2 = 9. 415 4
Now we have:
Period 0
3
Vu3 = 0
Vu2 = 0
Vu = 0.740 9
V = 3. 292 9
Vu2 d = 0
Vud = 1. 400 9
Vd = 5. 602 9
Vd2 u = 2. 648 7
Vd2 = 9. 415 4
Vd3 = 13. 584 3
Step 5
Cu Cd
Su Cd Sd Cu
B = erh
Su Sd
Su Sd
Period 0
63
2
(4, B)u2 = (0, 0)
Vu3 Vu2 d
00
= e0(1/3)
=0
3
2
Su Su d
74.6781 52.8140
Su3 Vu2 d Su2 dVu2 d
74.6781 (0) 52.8140 (0)
= erh
= e0.08(1/3)
=0
3
2
Su Su d
74.6781 52.8140
4u2 = eh
Bu2
Vu2 d Vd2 u
0 2. 648 7
= e0(1/3)
= 0.171 3
2
2
Su d Sd u
52.8140 37.3513
2
2
Su dVd2 u Sd Vu2 d
52.8140 (2. 648 7) 37.3513 (0)
= erh
= e0.08(1/3)
=
2
2
Su d Sd u
52.8140 37.3513
4ud = eh
Bud
8. 808 8
Vd2 u Vd3
2. 648 7 13. 584 3
= e0(1/3)
= 1.0
Su2 d Sd3
37.3513 26.4157
Su2 dVd3 Sd3 Vd2 u
37.3513 (13. 584 3) 26.4157 (2. 648 7)
= e0.08(1/3)
Bd2 = erh
=
Su2 d Sd3
37.3513 26.4157
38. 947 4
V 2 Vud
0 1. 400 9
= e0(1/3)
= 0.07 82
4u = eh u2
Su Sud
61.1491 43.2460
Su2 Vud SudVu2
61.1491 (1. 400 9) 43.2460 (0)
Bu = erh
= e0.08(1/3)
=
Su2 Sud
61.1491 43.2460
4. 659
4d2 = eh
Vud Vd2
1. 400 9 9. 415 4
= e0(1/3)
= 0.632 99
2
Sud Sd
43.2460 30.5846
2
SudVd2 Sd Vud
43.2460 (9. 415 4) 30.5846 (1. 400 9)
= e0.08(1/3)
Bd = eh
=
2
Sud Sd
43.2460 30.5846
28. 017 8
Vu Vd
0.740 9 5. 602 9
4 = eh
= e0(1/3)
= 0.331 66
Su Sd
50.0711 35.4114
Vu Vd
50.0711 (5. 602 9) 35.4114 (0.740 9)
B = erh
= e0.08(1/3)
= 16.
Su Sd
50.0711 35.4114
890 9
4d = eh
Please note that for an American option, you cant use the following approach to find the option price:
64
Node
Payo
Risk neutral prob of reaching this node
3
u3
Vu3 = 0
3u= 0.456 806
2
2
2
u d
Vu2 d = 0
3u d = 3 0.456 806 0.543 194
ud2
Vd2 u = 2. 648 7
3 u 2d = 3 (0.456 806) 0.543 1942
3
d
Vd3 = 13. 584 3
3d = 0.543 1943
= e0.08(1) 2. 648 7 3 0.456 806 0.543 1942 + 13. 584 3 0.543 1943 =
2. 998 5
This approach is wrong because it ignores the possibility that the American
option can be exercised early.
10.2.2
The price of an option on stock index can be calculated the same way as the
price of an option on a stock is calculated.
Example 10.2.11. Lets reproduce Derivatives Markets Figure 10.8. Here is
the recap of the information on an American call. The current stock index
is 110. The strike price K = 100. The annualized standard deviation of the
continuously compounded stock index return is = 30%. The continuously
compounded riskfree rate per year is r = 5%. The continuously compounded
dividend rate per year is = 3.5%.The option expiration date is T = 1 year.
Use a 3period binomial tree to calculate the option premium.
Solution.
Each period is h =
1
year long.
3
2
Vu2
Vu
V
Vud
65
3
Vu3 = max (0, 187.7471 100) = 87. 747 1
Vu2 d = max (0, 132.7789 100) = 32. 778 9
Vd
Vd2
3
Vu3 = 87. 747 1
Vd
Vd2 u = 0
Vd2 = 0
Vd3 = 0
Step 4
Calculate the value of the American call at Period 1 by taking
the greater of the backwardized value and the exercise value at each node.
Cu = erh (u Vu2 + d Vud ) = e0.05(1/3) (0.456 807 57. 101 3 + 0.543 193 14. 726 1) =
33. 520 02
66
Cd = erh ( u Vud + d Vd2 ) = e0.05(1/3) (0.456 807 14. 726 1 + 0.543 193 0) =
6. 615 8
EVu = max (0, Su K) = max (0, 131.4577 100) = 31. 457 7
EVd = max (0, Sd K) = max (0, 92.9699 100) = 0
Vu = max (Cu , EVu ) = max (33. 520 02, 31. 457 7) = 33. 520 02
Vd = max (Cd , EVd ) = max (6. 615 8, 0) = 6. 615 8
Now we have:
Period 0
3
Vu3 = 87. 747 1
Vd = 6. 615 8
Vd2 u = 0
Vd2 = 0
Vd3 = 0
Step 5
Calculate the value of the American call at Period 0 by taking
the greater of the backwardized value and the exercise value at each node.
C = erh ( u Vu + d Vd ) = e0.05(1/3) (0.456 807 33. 520 02 + 0.543 193 6. 615 8) =
18. 593 35
EV = max (0, S K) = max (0, 110 100) = 10
V = max (C, EV ) = max (18. 593 35, 10) = 18. 593 35
Step 6
Calculate the replicating portfolio
Our goal is to replicate the following values:
Period 0
1
2
3
Vu3 = 87. 747 1
Vd = 6. 615 8
Vd2 u = 0
Vd2 = 0
Vd3 = 0
Cu Cd
Su Sd
Period 0
4 = eh
4 = 0.690 923
B = erh
1
Su Cd Sd Cu
Su Sd
2
4u = 0.910 598
4d = 0.447 45
B = 57. 408 1
Bu = 86. 185 1
Bd = 34. 983 9
67
2
Bu2 = 98. 347 1
Bud = 77. 870 8
Bd2 = 0
Vu3 Vu2 d
87. 747 1 32. 778 9
= e0.035(1/3)
= 0.988 401
3
2
Su Su d
187.7471 132.7789
3
2
Su Vu2 d Su dVu3
187.7471 (32. 778 9) 132.7789 (87. 747 1)
= e0.05(1/3)
=
Bu2 = erh
3
2
Su Su d
187.7471 132.7789
98. 347 1
4u2 = eh
Vu2 d Vud2
32. 778 9 0
= e0.035(1/3)
= 0.988 401
Su2 d Sud2
187.7471 132.7789
3
2
Su dVu2 d Su dVu3
187.7471 (0) 132.7789 (32. 778 9)
= e0.05(1/3)
=
Bud = erh
Su3 Su2 d
187.7471 132.7789
77. 870 8
4ud = eh
Vd2 u Vd3
=0
Sd2 u Sd3
Sd2 uVd3 Sd3 Vd2 u
= erh
=0
Sd2 u Sd3
4d2 = eh
Bd2
Vu2 Vud
57. 101 3 14. 726 1
= e0.035(1/3)
= 0.910 598
2
Su Sud
157.1013 111.1055
2
Su Vud SudVu2
157.1013 (14. 726 1) 111.1055 (57. 101 3)
= e0.05(1/3)
=
Bu = erh
2
Su Sud
157.1013 111.1055
86. 185 1
4u = eh
Vud Vd2
14. 726 1 0
= e0.035(1/3)
= 0.447 45
Sud Sd2
111.1055 78.5763
2
Su Vud SudVu2
111.1055 (0) 78.5763 (14. 726 1)
= e0.05(1/3)
=
Bd = erh
Su2 Sud
111.1055 78.5763
34. 983 9
4d = eh
Vu Vd
33. 520 02 6. 615 8
= e0.035(1/3)
= 0.690 923
Su Sd
131.4577 92.9699
SuVd SdVu
131.4577 (6. 615 8) 92.9699 (33. 520 02)
B = erh
= e0.05(1/3)
=
Su Sd
131.4577 92.9699
57. 408 1
4 = eh
10.2.3
Options on currency
Now lets find the price of a European call option on 1. The underlying asset is
1. The option expires in h years. The current dollar value of the underlying is
68
h
Su
time 0
h
Cu = max (0, Su K)
C
Sd
Cd = max (0, Sd K)
To replicate the payo, at t = 0 well buy units of the underlying (i.e.
buy euros) and simultaneously invest B dollars in a savings account. Since
the underlying asset earns interest at a continuous interest , it will grow
into eh at T , which is worth $eh Su in the up state and $eh Sd in the
down state.
We want our replicating portfolio to match the option payo. So we have:
eh Su
Berh
Cu
4S
+ B
= C
eh Sd
Berh
Cd
t=0 t=h
t=0 t=h
t=0 t=h
eh Su + Berh = Cu
eh Sd + Berh = Cd
Solving these equations, we get:
Cu Cd
Su Sd
(10.26)
Su Cd Sd Cu
Su Sd
(10.27)
4 = eh
B = erh
Equation 10.26 and 10.27 are exactly the same as Equation 10.13 and 10.14.
This tells us that if we treat the currency as a stock and treat the euro return
as the stocks dividend rate, we can find the currency options price and the
replicating portfolios using all the formulas available for a stock option.
Example 10.2.12. Reproduce Derivatives Markets Figure 10.9. Here is the recap of the information on an American put option on 1. The current exchange
rate is S = $1.05/. The strike price is K = $1.10. The annualized standard
deviation of the continuously compounded return on dollars is = 10%. The
continuously compounded riskfree rate on dollars is r = 5% per year. The
continuously compounded return on euros is = 3.1% per year. The option
69
expiration date is T = 0.5 year. Use a 3period binomial tree to calculate the
option premium.
Solution.
The length period is h =
T
1
=
3
6
2
Vu2
Vu
V
Vud
Vd
Vd2
3
Vu3 = max (0, 1.10 1.2011) = 0
Vu2 d = max (0, 1.10 1.1070) = 0
Vd2 u = max (0, 1.10 1.0202) = 0.079 8
Vd3 = max (0, 1.10 0.9402 ) = 0.159 8
Step 3
Calculate the value of the American put at Period 2 by taking
the greater of the backwardized value and the exercise value at each node.
Cud = erh (u Vu2 d + d Vd2 u ) = e0.055(1/6) (0.489 795 0 + 0.510 205 0.079 8) =
0.04 034
Cd2 = erh ( ud Vd2 u + d Vd3 ) = e0.055(1/6) (0.489 795 0.079 8 + 0.510 205 0.159 8) =
0.119 5
70
3
Vu3 = 0
Vu2 = 0
Vu
V
Vu2 d = 0
Vud = 0.041 6
Vd
Vd2 u = 0.079 8
Vd2 = 0.124 6
Vd3 = 0.159 8
Step 4
Calculate the value of the American put at Period 1 by taking
the greater of the backwardized value and the exercise value at each node.
Cu = erh ( u Vu2 + d Vud ) = e0.055(1/6) (0.489 795 0 + 0.510 205 0.041 6) =
0.02 10
Cd = erh ( u Vud + d Vd2 ) = e0.055(1/6) (0.489 795 0.041 6 + 0.510 205 0.124 6) =
0.08 31
EVu = max (0, K Su ) = max (0, 1.10 1.0981) = 0.001 9
EVd = max (0, K Sd ) = max (0, 1.10 1.0120 ) = 0.088
Vu = max (Cu , EVu ) = max (0.02 10 , 0.001 9) = 0.021
Vd = max (Cd , EVd ) = max (0.088 , 0.021) = 0.088
Now we have:
Period 0
3
Vu3 = 0
Vu2 = 0
Vu = 0.021
V
Vu2 d = 0
Vud = 0.041 6
Vd = 0.088
Vd2 u = 0.079 8
Vd2 = 0.124 6
Vd3 = 0.159 8
Step 5
Calculate the value of the American put at Period 0 by taking
the greater of the backwardized value and the exercise value at each node.
71
3
Vu3 = 0
Vu2 = 0
Vu = 0.021
V = 0.054 7
Vu2 d = 0
Vud = 0.041 6
Vd = 0.088
Vd2 u = 0.079 8
Vd2 = 0.124 6
Vd3 = 0.159 8
4 = 0.7736
Period 0
4u = 0.4592
4d = 0.9948
4u2 = 0.0000
4ud = 0.9151
4d2 = 0.9948
2
Bu2 = $0.0000
Bu = $0.5253
B = $0.8669
Bud = $1.0089
Bd = $1.0900
Bd2 = $1.0900
10.2.4
Suppose we want to find the price of a European call option on a stock futures
contract. The underlying asset is futures. The option expires in h years. The
current price of the underlying asset is F0,h = F , where F0,h is the price of a
futures contract signed at t = 0 and expiring on date h.In h years the futures
price can go up to Fh,h = Fu = F u or go down to Fh,h = Fd = F d, where Fh,h
is the price of a futures contract signed at t = h and expiring on date h years
(i.e. expiring immediately). Fh,h is equal to the spot price Sh . The fact that
Fh,h can be either Fu or Fd is the same as fact that the stock price at t = h is
72
h
Fu = F u
time 0
h
Cu = max (0, Fu K)
C
Fd = F d
Cd = max (0, Fd K)
We form a replicating portfolio at t = 0 by entering futures contracts as
a buyer and simultaneously putting $B in the savings account. Assume that
no margin account is needed before one enters a futures contract. Then the
cost of entering a futures contracts is zero. At the contract expiration date h,
the futures contracts are settled in cash. If the futures price at expiration is
Fu > F , then the seller in the futures pays (Fu F ) to us, the buyer.
If on the other hand, the futures price at h is Fd < F , then we pay the seller
(F Fd ). Paying (F Fd ) is the same as receiving (Fd F ).
We assume that Fd < F < Fu holds so theres no arbitrage.
So the cash flow of the underlying asset (futures) is:
time 0
h
$ (Fu F )
$0
$ (Fd F )
We want the replicating portfolio
(Fu F )
0
+ B
(Fd F )
t=0 t=h
t=0
(Fu F ) + Berh = Cu
(Fd F ) + Berh = Cd
1d
u1
rh
Cu
+ Cd
B=e
ud
ud
1d
u1
+ Cd
V = 4 0 + B = B = erh Cu
ud
ud
4=
(10.28)
(10.29)
(10.30)
73
Define
u =
F Fd
1d
=
Fu Fd
ud
(10.31)
d =
Fu 1
u1
=
Fu Fd
ud
(10.32)
Then
V = erh (Cu u + Cd d )
(10.33)
Cu Cd
1d
u1
and that B = V = erh Cu
+ Cd
instead of
4 = er
Fu Fd
ud
ud
Su Cd Sd Cu
.
B = erh
Su Sd
How we can specify u or d?
u=
up price of Fh,h
F0,h
u=
S0 e(r)h+h
. In addition, F0,h = S0 e(r)h .
S0 e(r)h h
up price of Fh,h
S0 e(r)h+ h
h
u=
=
=
e
F0,h
S0 e(r)h
u = e
d = e
(10.34)
(10.35)
Equation 10.34 and 10.35 are the same as Equation 10.24 and 10.25 if we
set = r. We can use the stock options formula on u and d for futures options.
Tip 10.2.6. To find the price of a futures option, just use the price formula for
a stock option and set = r. However, remember that for a futures option, 4 =
Cu Cd
Cu Cd
1d
u1
instead of 4 = erh
and B = V = erh Cu
+ Cd
Fu Fd
Fu Fd
ud
ud
Su Cd Sd Cu
.
instead of B = erh
Su Sd
74
h
= e0.1 1/3
u = e(r)h+ h = e
= 1. 059 434
d = e(r)h h = e h = e0.1 1/3 = 0.943 900
e(r)h d
1d
1 0.943 900
u =
=
=
= 0.485 57
ud
ud
1. 059 434 0.943 900
d = 1 u = 1 0.485 57 = 0.514 43
Each period is h =
Step 1
Find the underlying asset price tree
Period 0
1
2
3
Su3 = 356.7330
Su2 = 336.7203
Su2 d = 317.8303
Su = 317.8303
F = 300
Sud = 300.0000
Sd2 u = 283.1700
Sd = 283.1700
2
Sd = 267.2842
Sd3 = 252.2895
Step 2
Period 0
2
Vu2
Vu
V
Vud
Vd
Vd2
3
Vu3 = max (0, 356.7330 300) = 56. 733
Vu2 d = max (0, 317.8303 300) = 17. 830 3
Vd2 u = max (0, 283.1700 300) = 0
Vd3 = max (0, 252.2895 300) = 0
Step 3
Calculate the value of the American call at Period 2 by taking
the greater of the backwardized value and the exercise value at each node.
The backwardized values at Period 2 are:
75
Cu2 = erh (u Vu3 + d Vu2 d ) = e0.05(1/3) (0.485 57 56. 733 + 0.514 43 17. 830 3) =
36. 113 3
Cud = erh (u Vu2 d + d Vd2 u ) = e0.05(1/3) (0.485 57 17. 830 3 + 0.514 43 0) =
8. 514 7
Cd2 = erh ( ud Vd2 u + d Vd3 ) = e0.05(1/3) (0.485 57 0 + 0.514 43 0) =
0
The exercise values at Period 2 are:
3
Vu3 = 56. 733
Vd
Vd2 u = 0
Vd2 = 0
Vd3 = 0
Step 4
Calculate the value of the American call at Period 1 by taking
the greater of the backwardized value and the exercise value at each node.
Cu = erh (u Vu2 + d Vud ) = e0.05(1/3) (0.485 57 36. 720 3 + 0.514 43 8. 514 7) =
21. 843 4
Cd = erh ( u Vud + d Vd2 ) = e0.05(1/3) (0.485 57 8. 514 7 + 0.514 43 0) =
4. 066 2
EVu = max (0, Su K) = max (0, 317.8303 300) = 17. 830 3
EVd = max (0, Sd K) = max (0, 283.1700 300) = 0
Vu = max (Cu , EVu ) = max (21. 843 4, 17. 830 3) = 21. 843 4
Vd = max (Cd , EVd ) = max (4. 066 2, 0) = 4. 066 2
Now we have:
76
3
Vu3 = 56. 733
Vd = 4. 066 2
Vd2 u = 0
Vd2 = 0
Vd3 = 0
Step 5
Calculate the value of the American call at Period 0 by taking
the greater of the backwardized value and the exercise value at each node.
C = erh ( u Vu + d Vd ) = e0.05(1/3) (0.485 57 21. 843 4 + 0.514 43 4. 066 2) =
12. 488 4
EV = max (0, S K) = max (0, 300 300) = 0
V = max (C, EV ) = max (12. 488 4, 0) = 12. 488 4
Now we have:
Period 0
1
2
3
Vu3 = 56. 733
Vu2 = 36. 720 3
Vu = 21. 843 4
Vu2 d = 17. 830 3
V = 12. 488 4
Vud = 8. 514 7
Vd = 4. 066 2
Vd2 u = 0
Vd2 = 0
Vd3 = 0
Next, we need to find the replicating portfolio. Our goal is to replicate the
following values:
Period 0
1
2
3
Vu3 = 56. 733
Vu2 = 36. 720 3
Vu = 21. 843 4
Vu2 d = 17. 830 3
V = 12. 488 4
Vud = 8. 514 7
Vd = 4. 066 2
Vd2 u = 0
Vd2 = 0
Vd3 = 0
Using Equation 10.29 and 10.29, you should get:
Period 0
1
2
4 = 0.5129
4u = 0.7681
4d = 0.2603
4u2 = 1
4ud = 0.5144
4d2 = 0
77
2
Bu2 = $36. 720 3
Bu = $21. 843 4
B = $12. 488 4
For example,
V 3 Vu2 d
56. 733 17. 830 3
4u2 = u
=
= 1.0
Su3 Su2 d
356.7330 317.8303
Bu2 = Vu2 = $36. 720 3
21. 843 4 4. 066 2
Vu Vd
=
= 0.512 9
Su Sd
317.8303 283.1700
B = V = $12. 488 4
I recommend that you reproduce my replicating portfolio in each node.
4=
Options on commodities
The textbook is brief on this topic. So you dont need to spend lot of time on it.
This is the main idea: we can price commodity options using the same framework
for pricing stock options if we can build a replicating portfolio using commodities
and bonds with zero transaction cost. In reality, its hard to build a replicating
portfolio using commodities. Unlike stocks, commodities such as corn may incur
storage cost or other cost. It may be impossible to short sell commodities.
As such, our ability to build a replicating portfolio is limited. However, if
we can build any replicating portfolios using commodities instantaneously and
eortlessly, commodity options and stock options are conceptually the same.
We can use the same framework to calculate the price of a commodity option
and the price of a stock option.
Example 10.2.14. Here is the information on an American call on a commodity. The current commodity price is 110. The strike price is K = 100.
The annualized standard deviation of the continuously compounded return on
the commodity is = 30%. The continuously compounded riskfree rate per
year is r = 5%. The continuously compounded lease rate of the commodity is
= 3.5% per year. The option expiration date is T = 1 year. Use a 3period
binomial tree to calculate the option premium. Assume that you can eortlessly
and instantaneously build any replicating portfolio using the commodity and the
bond.
We just treat the commodity as a stock. The lease rate = 3.5% is the same
as the stock dividend rate. We can use the framework for pricing stock options
to price this commodity option. The solution to this problem is in the textbook
Figure 10.8.
78
Options on bonds
The textbook points out two major dierences between bonds and stocks:
1. A bonds volatility decreases over time as the bond approaches its maturity. A stocks volatility doesnt have this pattern.
2. When pricing a stock option, we assume that the interest rate is constant
over time. The random variable is the stock price under a fixed interest.
However, when pricing a bond, we cant assume that the interest rate is
constant any more. If the interest rate is constant, then the bonds price
is known with 100% certainty. If the bond price doesnt change randomly,
an option on the bond has zero value. Who wants to buy a call or put on
an asset whose price is known with 100% certainty?
Because of these dierences, options on bonds should be treated dierently
from options on stocks. This is all you need to know about bond options right
now. Chapter 24 will cover more on options on bonds.
Chapter 11
textbook errata say the formulas work for an infinitelylived American call option.
79
80
The annual dividend you
gain if you exercise the call early is ST e 1 =
1
ST 1 + + 2 + ... 1 ST for a small . The annual interest earned on
2
1
r
K is K (e 1) = K 1 + r + r2 + ... 1 rK for a small r. If you early
2
exercise, youll pay K and receive ST ; during a year youll receive ST dividend
but you will lose rK interest. Hence early exercise is optimal if the annual
dividend exceeds the annual interest, ST > rK.
And its optimal to defer exercising a perpetual American call if the interest
savings on the strike price exceeds dividends lost:
ST < rK
Example 11.1.1. A perpetual American call option on a stock has a strike
price $50. The stock pays dividend at a continuously compounded rate of 8%
per year. The continuously compounded riskfree interest rate is 6% per year.
The volatility of the stock price is zero. When is it optimal to exercise this
American call option early? When is it optimal to defer exercise?
S > rK
rK
0.06 (50)
S>
=
= 37. 5
0.08
Once the stock price becomes greater than 37. 5, then its optimal to exercise
this perpetual American call option early.
If the stock price is less than 37. 5, then its optimal to defer exercising this
perpetual American call option early.
Please note that zero volatility doesnt mean that the stock price is a constant. It means that the stock price is known in advance with 100% certainty.
Next, the textbook says that the decision to exercise a perpetual American
call option early is complex if the volatility of the stock price is greater than
zero. In this case, the insurance in the call option is greater than zero. For
each nonzero volatility, theres a lowest stock price at which the early exercise
is optimal.
11.2
Risk neutral probability is explained in the previous chapter. Theres not much
new information about the risk neutral probability in this chapter. The key
point to remember is that risk neutral probability is a shortcut or a math trick
that enables us to quickly find the price of an option. We can use risk neutral
probability to find the correct price of an option whether consumers are really
risk neutral or not.
The risk neutral probability is similar to the moment generating function in
Exam P. The moment generating function (M GF ) is merely a math trick that
81
allows us to quickly find the mean and variance of a random variable (hence
the name "moment generating"). If we dont use GM F , we can still find the
mean and variance, but we have to work a lot harder. With the help of GM F ,
we can quickly find the mean and variance. Similarly, if we dont use the risk
neutral probability, we can still find the option price, but we have to work a lot
harder. Once we use riskneutral probability, we can quickly find an options
price. Risk neutral probability is merely a math risk.
By the way, one investment consultant told me that risk neutral probability
is often hard to nontechnical clients to understand. If you tell a nontechnical
client that an option price is calculated using riskneutral probability and that
the risk neutral probability not real, the client often immediately ask "So the
price you calculated is wrong then?" It may take the consultant a while to
explain why the risk neutral probability is not real yet the price is still correct.
11.2.1
Next, the textbook answers a frequently asked question: Can we calculate the
option price as the expected payo using real probabilities of the stock price?
The answer is "Yes if you know the discount rate."
Let p and q = 1p represent the real probability of stock going up and down
respectively. Let represent the real discount rate (instead of the riskfree rate,
which is used to discount payo in the risk neutral world).
uS with real probability p
S
t=0
t=0
e()h d
p=
S = eh puSeh + qdSeh
ud
p=
e()h d
ud
(11.1)
82
q=
u e()h
ud
(11.2)
Tip 11.2.1. If we set r = Equation 10.16 and 10.17 become Equation 11.1
and 11.2. So you just need to memorize Equation 10.16 and 10.17. To get the
formulas for the real probability, just set r = .
We can use the replicating portfolio to find . Suppose the replicating portfolio at t = 0 consists of shares of stock and putting $B in a savings account.
We already know that we can calculate and B using Equation 10.13 and
10.13.
At t = 0, the replicating portfolio is worth S + B. At t = h, the replicating
portfolio consists of eh shares of stock and $Berh in a savings account, which
is worth eh dS + Berh in the up state and eh dS Berh in the down state.
uSeh + Berh with real probability p
S + B
Value t = 0
The value at
value
at t = h discounted
at rate :
t = 0 is the expected
S + B = p uSeh + Berh + q dSeh + Berh eh
= Seh + Berh eh
eh =
S + B
Seh + Berh
(11.3)
S + B
(pCu + qCd )
Seh + Berh
Since C = S + B, we just need to prove that pCu + qCd = Seh + Berh .
C = eh (pCu + qCd ) =
pCu + qCd
e()h d
u e()h
=
Cu +
C
ud
ud
(r)h
()h
e
e(r)h e()h
e
u e(r)h
d
e(r)h
=
Cu +
Cu +
Cd +
Cd
ud
ud
ud
ud
(r)h
e
d
u e(r)h
e()h e(r)h
=
Cu +
Cd +
(Cu Cd )
ud
ud
ud
According to Equation 10.15:
u e(r)h
e(r)h d
Cu +
Cd = erh (S + B)
ud
ud
83
Cu Cd
= eh S
ud
pCu + qCd
S + B
(pCu + qCd ) = S + B
Seh + Berh
The above derivation tell us that
C=
Real probabilities lead to the same answer as the risk neutral probability
Any consistent pair of (, ) will produce the correct answer. The above
derivation doesnt require that has to be reasonable or precise. Any
will generate a corresponding . Together, and will produce the option
price correctly.
The simplest calculation is to set = = r. Setting = = r means
using risk neutral probabilities.
Example 11.2.1. Reproduce the textbook Figure 11.3 (which is the same as the
textbook Figure 10.5). A European call option has strike price K = 40. The
current price is S = 41. The annualized standard deviation of the continuously
compounded stock return is = 30%. The continuously compounded riskfree
rate per year is r = 8%. The continuously compounded dividend rate per year
is = 0%. The continuously compounded expected return on the stock per year
is = 15%. The option expiration date is T = 1 year. Use a 1period binomial
tree and real probabilities to calculate the option premium.
Solution.
C =?
Cu = max (0, 59. 954 40) = 19. 954 with real probability p
t=0
84
q = 1 p = 1 0.544 6 = 0.455 4
discounting rate:
S + B = Seh + Berh eh
(S + B) eh = Seh + Berh
(0.737 6 41 22. 403 6) e(1) = 0.737 6 41e0.15(1) 22. 403 6e0.08(1)
e(1) = 1. 386
= 0.326 4
Calculate the option price:
C = eh (pCu + qCd ) = e0.326 4(1) (0.544 6 19. 954 + 0.455 4 0) = 7. 84
This option price is the same as the price in the textbook Figure 10.3.
Example 11.2.2. Reproduce the textbook Figure 11.4 (the risk neutral solution
to an otherwise identical European call is in the textbook Figure 10.5). Here is
the recap of the information on an American call. The current stock price is 41.
The strike price K = 40. The annualized standard deviation of the continuously
compounded stock return is = 30%. The continuously compounded riskfree
rate per year is r = 8%. The continuously compounded expected return on the
stock per year is = 15%.The continuously compounded dividend rate per year
is = 0%.The option expiration date is T = 1 year. Use a 3period binomial
tree and real probabilities to calculate the option premium.
Solution.
Each period is h =
1
year long.
3
85
The replicating portfolios are copied over from the textbook Figure 10.5.
Period 0
2
(4, B)u2 = (1, 38. 947 4)
(4, B)ud = (0.828 7, 30. 138 6)
(4, B)d2 = (0, 0)
In addition, we need to calculate the discount rate for each node. We put
the stock price table and the replicating portfolio table side by side:
Stock price
(4, B)
Period 0
1
2
3
Period 0
1
74.6781
61.1491
50.0711
52.8140
(0.921 8, 33. 263 6)
41
43.2460
(0.706 3, 21. 885 2)
35.4114
37.3513
(0.450 1, 13. 405 2)
30.5846
26.4157
0:
Calculate the common discounting factor Node u Node 0 and Node d Node
e(1/3) =
S + B
0.706 3 (41) 21. 885 2
=
=
Seh + Berh
0.706 3 (41) e0.15(1/3) 21. 885 2e0.08(1/3)
0.887 87
e(1/3) = 0.887 87
= 0.356 8
S + B
0.921 8 1 (50.0711) 33. 263 6
=
=
h
rh
Se + Be
0.921 8 (50.0711) e0.15(1/3) 33. 263 6 2e0.08(1/3)
0.897 84
e(1/3) = 0.897 84
= 0.323 3
S + B
0.450 1 (35.4114) 13. 405 2
=
=
Seh + Berh
0.450 1 (35.4114) e0.15(1/3) 13. 405 2 64e0.08(1/3)
0.847 79
e(1/3) = 0.847 79
= 0.495 4
2
(1, 38. 947 4)
(0.828 7, 30. 138 6)
(0, 0)
86
S + B
1 (61.1491) 38. 947 4
=
=
Seh + Berh
1 (61.1491) e0.15(1/3) 38. 947 4e0.08(1/3)
e(1/3) =
0.914 24
e(1/3) = 0.914 24
= 0.2690
e(1/3) =
S + B
0.828 7 (43.2460) 30. 138 6
=
=
h
rh
Se + Be
0.828 7 (43.2460) e0.15(1/3) 30. 138 6e0.08(1/3)
0.847 83
e(1/3) = 0.847 83
= 0.495 2
Calculate the common discounting factor Node ud2 Node d2 and Node
d Node d2 :
3
S + B
0 (30.5846) 0
=
= N/A
Seh + Berh
0 (30.5846) e0.15(1/3) 0 6e0.08(1/3)
= N/A
= N/A
e(1/3) =
e(1/3)
2
= 0.2690
3
Cu3 = max (0, 74.6781 40) = 34. 678 1
= 0.495 2
= N/A
= 0.323 3
= 0.356 8
= 0.495 4
3
Su3 = 74.6781
Su2 = 61.1491
Su2 d = 52.8140
Su = 50.0711
S = 41
Sud = 43.2460
Sd2 u = 37.3513
Sd = 35.4114
2
Sd = 30.5846
Sd3 = 26.4157
3
Vu3 = 34. 678 1
= 0.495 4
Vud2 = 0
Vd2 = 0
Vd3 = 0
Similarly,
Cu = (22. 201 2 0.524 6 + 5. 699 4 0.475 4) e0.323 3(1/3) = 12. 889 6
EVu = max (0, 50.0711 40) = 10. 071 1
Vu = max (12. 889 6, 10. 071 1) = 12. 889 6
Cd = (5. 699 4 0.524 6 + 0 0.475 4) e0.495 4(1/3) = 2. 534 8
EVd = max (0, 35.4114 40) = 0
Vd = max (2. 534 8, 0) = 2. 534 8
Now we have:
Period 0
1
3
Vu3 = 34. 678 1
Vd = 2. 534 8
Vud2 = 0
Vd2 = 0
Vd3 = 0
Finally,
C = (12. 889 6 0.524 6 + 2. 534 8 0.475 4) e0.356 8(1/3) = 7. 0734
EV = max (0, 41 40) = 1
87
88
3
Vu3 = 34. 678 1
Vd = 2. 534 8
Vud2 = 0
Vd2 = 0
Vd3 = 0
Tip 11.2.2. Real probability pricing requires intensive calculation. Not only
do we need to find the real probability of up and down, we also need to find
the replicating portfolio at each node. In contrast, in risk neutral pricing, we
either use risk neutral probabilities or use replicating portfolio but not both. In
comparison, risk neutral pricing is more ecient than real probability pricing.
11.2.2
89
To apply the random walk model to stock prices, we can use Yi to represent
the price movement during an interval of time. We can use Zn to represent the
ending price of a stock after n equal intervals.
To get a good feel of the random variable Zn , check out the simulation of
the random walk model at
http://math.furman.edu/~dcs/java/rw.htmll
Just type in the url in your web browser.
Modeling stock as a random walk
There are 3 problems if we use Zn = Y1 + Y2 + ... + Yn to model the price of a
stock:
1. Zn can be negative yet the price of a stock cant be negative.
2. The incremental change of any stock price Yi is either 1 or 1. Though
$1 change might be OK for modeling the change of a low priced stock, it
may be inappropriate to model the change of a high priced stock.
3. The expected return on stock whose price following a random walk is zero,
that is E (Zn ) = nE (Y ) = n 0 = 0. However, stock on average should
have a positive return.
Continuously compounded returns
Let rt,t+h represent the continuously compounded return earned during the time
interval [t, t + h]. Let St and St+h represent the stock price at time t and t + h
respectively. Then
St+h
St+h
St ert,t+h = St+h
ert,t+h =
rt,t+h = ln
St
St
Continuously compounded returns are additive. Consider the time interval
[t, t + nh]. We have:
St+nh
St+h
St+2h
St+3h
St+nh
ert,t+nh =
=
...
St
St
St+h
St+2h
St+(n1)h
= ert,t+h ert+h,t+2h ert+2h,t+3h ...ert+(n1)h,t+nh
rt,t+nh = rt,t+h +rt+h,t+2h +rt+2h,t+3h + ... + rt+(n1)h,t+nh
Continuously compounded returns can be negative. Even if r < 0, we still
have er > 0. Thus if ln S follows a random walk, S cant be negative.
Standard deviation of returns
The annual return is the sum of the returns in each of the 12 months:
rannual = rJan + rF eb + ... + rDec
Assume each monthly return is independent identical distributed with common variance 2Monthly . Let 2 represent the variance of the return over 1 year
period.
90
2 = 122M onthly
Monthly =
12
Suppose we split one year into n intervals with each interval being h long (so
nh = 1). Let h represent the standard deviation of the return over an interval
of h long, then
rannual = r0,h + rh,2h + r2h,3h
+ ... + r(n1)h,nh
2
= V ar (rannual ) = V ar r0,h + rh,2h + r2h,3h + ... + r(n1)h,nh = n 2h
h =
n
However, nh = 1. So we have: h = = h
n
Binomial model
(r)h+ h
(r)h h
and d = e
is equivalent
ing so is reasonable. Setting u = e
to setting
rt,t+h = (r ) h h
(11.4)
Lets see why Equation 11.4 solves the three problems in the random walk
model:
1. Even when rt,t+h is negative, the stock price St+h is always positive.
2. The change
priceiis proportional to the stock price. S = St+h
h in stock
(r)h h
1 .
St = St e
3. The expected return during [t, t + h] is largely driven by the constant term
(r ) h. Hence the expected return is no long always zero.
Alternative binomial tree
The CoxRossRubinstein binomial tree
u = e
d = e
(11.5)
(11.6)
2
d = e(r0.5 )h
(11.7)
(11.8)
11.2.3
91
Formulas
n+1
number of stock prices observed: S0 , S1 ,...,Sn1 ,and Sn
n
number of stock returns observed. Continuously compounded
return per period is estimated as
S1
S2
S3
Sn
r1 = ln
r2 = ln
r3 = ln
...rn = ln
S0
S1
S2
Sn1
Since our focus is stock returns not stock price, the number of observations
is n (i.e. the number of stock prices observed minus one). Remember this point.
The expected return is:
r1 + r2 + ... + rn
r=
n
The v
estimated standard deviation is:
2
u
2
2
u r
r
+
r
r
+
...
+
r
r
1
2
n
t
=
n1
Example 11.2.3. Reproduce the textbook Figure 11.1 and estimate the standard
deviation of the continuously compounded return per year earned by S&P 50
index.
St
2
Week S&P 500 price rt =ln
rt r
St1
0
829.85
1
804.19
0.0314
0.001846
2
874.02
0.0833
0.005143
3
869.95
0.0047
0.000263
4
880.9
0.0125
0.000001
5
865.99
0.0171
0.000819
6
879.91
0.0159
0.000019
7
919.02
0.0435
0.001020
8
916.92
0.0023
0.000191
9
929.62
0.0138
0.000005
10
939.28
0.0103
0.000001
11
923.42
0.0170
0.000817
12
953.22
0.0318
0.000409
Total
0.1386
0.010534
The expected continuously compounded return per week is estimated as:
r1 + r2 + ... + r12
0.1386
r=
=
= 0.011 55
n
12
The standard deviation of the continuously compounded return per week is
estimated as:
92
0.010534
= 0.03 095
12 1
1 Year = 52 Weeks
Let Y represent the continuously compounded return per year and Xi represent the continuously compounded return earned in the ith week.
Then Y = X1 + X2 + ... + X52
Where X1 , X2 , ..., X52 are assumed to be independent identically distributed.
V ar (Y) = V ar (X
1 + X2 + ... + X52 ) = 52V ar (X)
Y = 52 Y = 52 (0.03 095) = 0.223 18
Please note my calculation was done using Excel. If you cant perfectly
reproduce my numbers, thats fine.
By the way, in Excel, the formula for the mean is AVERAGE; the formula
for the sample variance is VAR
This is how to use Excel to calculate the continuously compounded return
per year earned by S&P 500 index. Suppose the stock prices are entered in
Column C (from C3 to C15) and the weekly returns are calculated in Column
D (from D4 to D15).
1
Column B
Column C
Week
Column D
St
rt =ln
St1
3
0
829.85
4
1
804.19
0.0314
5
2
874.02
0.0833
6
3
869.95
0.0047
7
4
880.9
0.0125
8
5
865.99
0.0171
9
6
879.91
0.0159
10
7
919.02
0.0435
11
8
916.92
0.0023
12
9
929.62
0.0138
13
10
939.28
0.0103
14
11
923.42
0.0170
15
12
953.22
0.0318
The expected continuously compounded return per week is:
r =AVERAGE(D4:D15)= 0.01155006
The sample variance of the continuously compounded return per week is
estimated as:
2
=VAR(D4:D15)= 0.000957682
= 0.000957682 = 0.030946434
Make sure you dont use the population variance formula:
VARP(D4:D15)= 0.000877875
93
2
2
2
r1 r + r2 r + ... + rn r
V ARP =
n
While VAR iscalculated
as
2
2
2
r1 r + r2 r + ... + rn r
V AR =
n1
So V AR = V ARP
n
n1
94
0
1
2
3
4
5
6
7
8
9
10
11
12
77.73
75.18
82
81.55
81.46
78.71
82.88
85.75
84.9
86.68
88.7
86.18
87.57
rt =ln
St
St1
0.0334
0.0868
0.0055
0.0011
0.0343
0.0516
0.0340
0.0100
0.0207
0.0230
0.0288
0.0160
95
52 (0.0365) = 0.263 2
11.3
Stu = St erh D e h
Std = St erh D e h
(11.9)
(11.10)
To find the price of the European option, we calculate the cost of the replicating portfolio. We have two assets: the stock and a savings account. The
savings account is the same as a zerocoupon bond. At time t + h, the stock
price is Sh ; the bond price is Bt+h . The bond price is deterministic:
Bt+h = erh
Bt = 1
The stock
( price at t + h is stochastic:
Stu = St erh D e h
St+h =
Std = St erh D e h
So at t + h the stock price either goes up to Stu ("up state") or goes down
to Std ("down state").
96
Std
Time t + h
4 (Stu + D)
4 Std + D
t+h
Berh
+
Cu
=
Berh
t+h
C
t
Cd
t+h
4 (Stu + D) + Berh = Cu
4 Std + D + Berh = Cd
Solving these equations, we get:
4=
B = erh
11.3.1
Cu Cd
Stu Std
Stu Cd Std Cu
4D
Stu Std
(11.11)
(11.12)
One major problem with the stock price tree using Equation 11.9 and 11.10 is
that the tree doesnt complete recombine after the discrete dividend.
Example 11.3.1. Reproduce the textbook Figure 11.1 but just for the 2 periods.
Show that the stock price tree doesnt recombine at Period 2. This is the recap
of the information. The current stock price is 41. The stock pays a dividend
during Period 1 and Period 2. The future value of the dividend accumulated
at the riskfree interest rate r from Period 1 to Period 2 is 5. Other data are:
r = 0.08, = 0.3, t = 1,and h = 1/3.
Stu
2
Stuu = 55. 203
= 50. 071
Stud
97
3
Stuuu = 67. 417 15
Stuud = Studu = 47. 678 91
= 39. 041
Studd = 33. 719 59
St = 41
Std
= 26. 380
Stddd = 22. 783 97
Stuu = Stu erh D e h = 50. 071 09e0.081/3 5 e0.3 1/3 = 55. 203 57
Stud = Stu erh D e h = 50. 071 09e0.081/3 5 e0.3 1/3 = 39. 041 20
Stdu = Std erh D e h = 35. 411 39e0.081/3 5 e0.3 1/3 = 37. 300 47
Stdd = Std erh D e h = 35. 411 39e0.081/3 5 e0.3 1/3 = 26. 379 73
Now you see that Stud 6= Stdu , . The tree doesnt recombine.
No dividend is paid during
Period 2 and Period 3.
Stuuu = Stuu erh D e h = 55. 203 57e0.081/3 0 e0.3 1/3 = 67. 417 15
Studu = Stud erh D e h = 39. 041 20e0.081/3 0 e0.3 1/3 = 47. 678 91
Stduu = Stdu erh D e h = 37. 300 47e0.081/3 0 e0.3 1/3 = 45. 553 05
Stddu = Stdd erh D e h = 26. 379 73e0.081/3 0 e0.3 1/3 = 32. 216 14
98
Stu = St erh 0 e h = Sterh e h
Stuu = Stu erh 5 e h = St erh e h 5 e h
= St erh e h 5 erh
Studu = Stud erh 0 e h = St erh e h 5 e h erh 0 e h
= St erh e h 5 erh
Clearly, Stuud = Studu . Similarly, you can verify for yourself that Stdud =
Stddu .
In this problem, Period 2 had 4 prices. If the stock pays continuous dividend,
Period 2 will have only 3 prices.
Similarly, in this problem, Period 3 has 5 distinct prices. In contrast, if the
stock pays continuous dividend, Period 3 will have only 4 prices.
In addition to noncombining stock prices, the above method may produce
negative stock prices.
11.3.2
Schroder presents a method that overcomes the two shortcomings of the above
method.
This is the idea behind Schroders method. Instead of directly building a
stock price tree (which proves to be noncombining after the discrete dividend is
paid), well build a tree of a series of prepaid forward prices on the same stock.
Hopefully, the prepaid forward
price
u
P
Ft+h,T
d
P
Ft+h,T
P
Ft+2h,T
uu
ud
P
Ft+2h,T
dd
P
Ft+2h,T
P
Ft,T
is the prepaid price of a forward contract signed at t expiring on date
T
99
P
Ft+h,T
is the prepaid price of a forward contract signed at t + h expiring
on date T
P
Ft+2h,T
is the prepaid price of a forward contract signed at t + 2h expiring
on date T
Der(TD t) if TD t
P
St = Ft,T +
0
if TD < t
Similarly,
Der(TD th)
= St+h P Vt+h (D) = St+h
0
r(TD th)
if TD t + h
De
P
+
St+h = Ft+h,T
0
if TD < t + h
P
Ft+h,T
if TD t + h
if TD < t + h
So theres a onetoone mapping between the prepaid forward price and the
stock price.
Next, lets find out how to build a prepaid forward price tree. We need to
know how the prepaid forward price changes over time. Suppose today is time
t . At t + h we enter into a forward contract agreeing to buy a stock at date T
where T > t + h. The stock will pay dividend D in date TD where t < TD < T .
If the stock volatility is zero (meaning that the future stock price is known today
with 100% certainty), then the price of the prepaid forward contact at t + h is
P
P rh
Ft+h,T
= Ft,T
e
P
Ft+h,T
P
Ft,T
= erh
(11.13)
(11.14)
P
at t, well receive one stock
This is why Equation 11.13 holds. If we pay Ft,T
P
P
at T . This gives us Ft,T
= ST er(T t) . Similarly, if we pay Ft+h,T
at t + h, well
100
P
also receive a stock at T . This gives us Ft+h,T
= ST er(T th) . Then Equation
11.13 holds.
Now suppose that the forward price has a volatility of F per year. Then
its reasonable
( to assume that
P rh h
P
F
e
e
= Ft,T
u
in the up state
t,T
P
Ft+h,T
=
P rh h
P
Ft,T
e e
= Ft,T
d in the down state
where u = erh+F
P
Similarly, Ft+2h,T
and d = erhF h
P
Ft+h,T
u
in the up state
=
P
d in the down state
Ft+h,T
u
P
P
= Ft,T
u
Ft+h,T
ud
du
P
P
P
P
Ft,T
= Ft+2h,T
= Ft,T
ud
Ft+2h,T
d
P
P
= Ft,T
d
Ft+h,T
dd
P
P 2
= Ft,T
d
Ft+2h,T
F = S
Once we have the prepaid forward price tree, well transform it into the stock
price tree:
P
St = Ft,T
+ P Vt (D)
u
u
P
St+h
= Ft+h,T
+ P Vt+h (D)
d
d
P
St+h
= Ft+h,T
+ P Vt+h (D)
uu
uu
P
St+2h
= Ft+2h,T
+ P Vt+2h (D)
ud
ud
P
St+2h
= Ft+2h,T
+ P Vt+2h (D)
dd
dd
P
St+2h
= Ft+2h,T
+ P Vt+2h (D)
Example 11.3.2. Let s reproduce the textbook Figure 11.11. Here is the recap
of the information on an American call option. The stock pays a dividend of $5
in 8 months. Current stock price is $41. The strike price K = $40. The stock
volatility is S = 0.3. The continuously compounded riskfree rate is r = 0.08.
The option expires in T = 1 year. Use a 3period binomial tree to calculate the
option price.
F = S
St
P
Ft,T
P
Ft,T
= St P Vt (D) = 41 5e0.08(8/12) = 36. 26
St
41
= 0.339 2
F = S P = 0.3
36. 26
Ft,T
P
Ft,T
u
P
Ft+h,T
d
P
Ft+h,T
uu
P
Ft+2h,T
ud
P
Ft+2h,T
dd
P
Ft+2h,T
uuu
P
Ft+3h,T
uud
P
Ft+3h,T
udd
P
Ft+3h,T
ddd
P
Ft+3h,T
u
P
P
= Ft,T
u = 36. 26 1. 249 20 = 45. 296
Ft+h,T
d
P
P
Ft+h,T
= Ft,T
d = 36. 26 0.844 36 = 30. 616
uu
u
P
P
P
= Ft,T
u2 = Ft+h,T
u = 45. 296 1. 249 20 = 56. 584
Ft+2h,T
ud
u
P
P
Ft+2h,T
= Ft+h,T
d = 45. 296 0.844 36 = 38. 246
dd
d
P
P 2
P
Ft+3h,T
= Ft,T
d = Ft+h,T
d = 30. 616 0.844 36 = 25. 851
uuu
uu
P
P
= Ft+2h,T
u = 56. 584 1. 249 20 = 70. 685
Ft+3h,T
uud
uu
P
P
Ft+3h,T
= Ft+2h,T
d = 56. 584 0.844 36 = 47. 777
udd
ud
P
P
Ft+3h,T
= Ft+2h,T
d = 38. 246 0.844 36 = 32. 293
ddd
dd
P
P
Ft+3h,T
= Ft+2h,T
d = 25. 851 0.844 36 = 21. 828
Prepaid forward price tree:
101
102
47. 777
38. 246
30. 616
32. 293
25. 851
21. 828
Next, we change the prepaid forward price tree into a stock price tree. The
onetoone mapping between
the prepaid forward price and the stock price is
r(TD tt)
if TD t + t
De
P
+
St+t = Ft+t,T
0
if TD < t + t
where 0 t T t
uuu
St+3/3
uu
St+2/3
uud
St+1/3
St+3/3
ud
St
St+2/3
udd
St+1/3
St+3/3
dd
St+2/3
ddd
St+3/3
8
+ t, D = 5,T = t + 1
12
r(TD t)
+ De
= 36. 26 + 5e0.08(8/12) = 41. 000
In this problem, TD =
P
St = Ft,T
u
u P
d P
St+1/3 = Ft+1/3,T
+ Der(TD t1/3) = 30. 616 + 5e0.08(8/121/3)
35. 484
uu
uu P
= Ft+2/3,T
+Der(TD t2/3) = 56. 584+5e0.08(8/122/3)
St+2/3
61. 584
ud
ud P
St+2/3
= Ft+2/3,T
+Der(TD t2/3) = 38. 246+5e0.08(8/122/3)
43. 246
dd
dd P
St+2/3
= Ft+2/3,T
+ Der(TD t2/3) = 25. 851 + 5e0.08(8/122/3)
30. 851
uuu
uuu P
= Ft+3/3,T
= 70. 685 (because TD < t + 3/3)
St+3/3
Similarly,
uud
uud P
St+3/3
= Ft+3/3,T
= 47. 777
udd
udd
P
St+3/3
= Ft+3/3,T
= 32. 293
ddd
ddd
P
St+3/3
= Ft+3/3,T
= 21. 828
=
=
=
=
=
103
47. 777
43. 246
35. 484
32. 293
30. 851
21. 828
After getting the stock price tree, we calculate the price of the American
call option as usual. We work backward from right to left. At each node, we
compare the backwardized value with the exercise value, taking the maximum
of the two.
The risk neutral probabilities are:
erh d
e0.08(1/3) 0.844 36
u =
=
= 0.451 2
ud
1. 249 20 0.844 36
d = 1 0.451 2 = 0.548 8
Payo tree:
Vuuu = 30. 685
Vuu = 21. 584
Vu = 11. 308
V = 5.770
Vuud = 7. 777
Vud = 3. 417
Vd = 1. 501
Vudd = 0
Vdd = 0
Vddd = 0
104
Chapter 12
BlackScholes
Except the option Greeks and the barrier option price formula, this chapter is
an easy read.
12.1
12.1.1
(12.1)
S
1 2
ln
+ r+ T
K
2
d1 =
T
d2 = d1 T
(12.2)
(12.3)
(12.4)
106
Example 12.1.1. Reproduce the textbook example 12.1. This is the recap of
the information. S = 41, K = 40,r = 0.08, = 0.3, T = 0.25 (i.e. 3 months),
and = 0. Calculate the price of the price of a European call option.
S
1 2
ln
+ r+ T
K
2
d1 =
T
1
41
+ 0.08 0 + 0.32 0.25
ln
40
2
= 0.3730
=
0.3 0.25
Example 12.1.2. Reproduce the textbook example 12.2. This is the recap of
the information. S = 41, K = 40,r = 0.08, = 0.3, T = 0.25 (i.e. 3 months),
and = 0. Calculate the price of the price of a European put option.
N (d1 ) = 1 N (d1 ) = 1 0.645 4 = 0.354 6
N (d2 ) = 1 N (d2 ) = 1 0.588 2 = 0.411 8
P = 41e0(0.25) 0.354 6 + 40e0.08(0.25) 0.411 8 = 1. 607
12.1.2
107
12.2
12.2.1
P
The prepaid forward price for the stock is: F0,T
(S) = SeT
P
The prepaid forward price for the strike asset is: F0,T
(K) = P V (K) =
rT
Ke
Define V (T ) = T
The price of a European call option in terms of repaid forward is:
P
P
P
P
(S) , F0,T
(K) , V (T ) = F0,T
(S) N (d1 ) F0,T
(K) N (d2 )
C F0,T
(12.5)
P
P
P
P F0,T
(S) , F0,T
(K) , V (T ) = F0,T
(S) N (d1 ) + F0,T
(K) N (d2 ) (12.6)
ln
d1 =
P
F0,T
(S)
P
F0,T
1
+ V 2 (T )
(K) 2
V (T )
d2 = d1 V (T )
(12.7)
(12.8)
108
12.2.2
When the stock pays discrete dividends, the prepaid forward price is:
P
(S) = S P V0,T (Div)
F0,T
P
(S) in Equation 12.5 and 12.6, you should get the price of the
Apply F0,T
European call and put where the stock pays discrete dividends.
Example 12.2.1. Reproduce the textbook example 12.3. Here is the recap of
the information. S = 41, K = 40, = 0.3, r = 0.08, and T = 0.25 (i.e. 3
months). The stock pays dividend of 3 in 1 month, but makes no other payouts
over the life of the option (so = 0). Calculate the price of the European call
and put.
P
(S) = S P V0,T (Div) = 41 3e(0.08)1/12 = 38. 020
F0,T
P
(K) = P V (K) = KerT = 40e0.08(0.25) = 39. 208
F0,T
12.2.3
Options on currencies
Notation
x, the current dollar value of 1
K, the strike price in dollars of 1
r, the continuously compounded riskfree rate earned by $1
rf , the continuously compounded riskfree rate earned by 1
, the annualized standard deviation of the continuously compounded
return on dollars
T , the option expiration date
C (x0 , K, , r, T, rf ), the price of a European call option with parameters
(x0 , K, , r, T, rf )
109
(12.9)
1
x
+ r rf + 2 T
K
2
d2 = d1 T
(12.10)
(12.11)
(12.12)
Tip 12.2.1. For currency options, just set S = x and = rf and apply the
BlackScholes formulas on European call and put. The same thing happened in
Equation 10.26 and 10.27.
Example 12.2.2. Reproduce the textbook example 12.4. Here is the recap of the
information. The current dollar price of 1 is $0.92. The strike dollar price of
1 is $0.9. The annualized standard deviation of the continuously compounded
return on dollars is = 0.1. The continuously compounded riskfree rate earned
by dollars is r = 6%. The the continuously compounded riskfree rate earned
by 1 is rf = 3.2%. The option expires in 1 year. Calculate the price of the
European call and put on 1.
1 2
x
+ r rf + T
ln
K
2
d1 =
0.92
1
ln
+ 0.06 0.032 + 0.12 1
0.9
2
=
= 0.549 8
0.1 1
110
12.2.4
Options on futures
For a futures contract, the prepaid price is just the present value of the futures
P
P
price. Set F0,T
(F ) = F erT and F0,T
(K) = KerT , we get:
ln
d1 =
P
F0,T
(F )
P
F0,T
1
+ 2T
(K) 2
(12.13)
d2 = d1 T
(12.14)
(12.15)
(12.16)
Example 12.2.3. Reproduce the textbook example 12.5. Here is the recap of
the information about the European option on a 1year futures contract. The
current futures price for natural gas is $2.10. The strike price is K = 2.10. The
volatility is = 0.25. r = 0.055, T = 1. Calculate the price of the European
call and put.
2.10 1
F
1
ln
+ 2T
+ 0.252 (1)
K
2
2.10
2
=
d1 =
= 0.125
0.25 1
T
d2 = d1 T = 0.125 0.25 1 = 0.125
N (d2 ) = 0.450 3
N (d1 ) = 0.549 7
N (d1 ) = 1 N (d1 ) = 1 0.549 7 = 0.450 3
N (d2 ) = 1 N (d2 ) = 1 0.450 3 = 0.549 7
C = F erT N (d1 ) KerT N (d2 )
= 2.10e0.055(1) 0.549 7 2.10e0.055(1) 0.450 3 = 0.197 6
P = F erT N (d1 ) + KerT N (dd2 )
= 2.10e0.055(1) 0.450 3 + 2.10e0.055(1) 0.549 7 = 0.197 6
ln
12.3
The learning objectives in the SOAs syllabus is to Interpret the option Greeks.
The learning objective in CAS Exam 3 Financial Economics is to Interpret
the option Greeks and elasticity measures. How to derive the option Greeks
is explained in the textbook Appendix 12.B, but Appendix 21.B is excluded
from the SOA MFE and CAS FE. So you might want to focus on the learning
objective of the exam.
Of all the Greeks, delta and gamma are the most important.
12.3.1
111
Delta
Cu Cd
when we try to find the replicating
Su Sd
portfolio of a European call or put. Its the number of stocks you need to
own at time zero to replicate the discrete payo of a European call or put at
C
C
. Here =
expiration date T . If the payo is continuous, then =
S
S
is the number of stocks you need to have now to replicate the payo of the
next instant (i.e. the payo one moment later). The European call price is
C
C = SeT N (d1 ) KerT N (d2 ) and the delta for a call is call =
=
S
eT N (d1 ).
One less visible thing to know is that d1 is also a function of S. So itll tame
C
some work to derive call =
= eT N (d1 ). One naive approach is treat
S
C
N (d1 ) as a constant and get call =
= eT N (d1 ). Interestingly, this gives
S
the correct answer!
Since deriving delta is not on the syllabus, you dont need to go through the
C
= eT N (d1 ). Just memorize that call =
messy math and prove call =
S
eT N (d1 ) for a European call and put = eT N (d1 ) = eT [1 N (d1 )] =
call eT for a European put.
Other results you might want to memorize:
0 call 1
1 put 0
Delta . You already see =
Example 12.3.1. Calculate the delta of the following European call and put.
The information is: S = 25,K = 20, = 0.15, r = 6%, = 2%, and T = 1
year.
25
S
1 2
1
2
ln
ln
+ r+ T
+ 0.06 0.02 + 0.15 1
K
2
20
2
=
=
d1 =
0.15 1
T
1. 829 3
N (d1 ) = 0.966 3
N (d1 ) = 1 0.966 3 = 0.033 7
call = eT N (d1 ) = e0.02(1) 0.966 3 = 0.947 2
put = eT N (d1 ) = e0.02(1) 0.033 7 = 0.03 30
12.3.2
Gamma
2C
=
=
S
S 2
If gamma is too large a small change in stock price will cause a big change
in . The bigger , the more often you need to adjust your holding of the
underlying stocks.
112
12.3.3
put
call
=
or call = put .
S
S
Vega
Vega is the change of option price for 1% change of stock volatility (you can
think that the letter V stands for volatility).
C
V ega =
100
12.3.4
Theta
Theta is the change of option price regarding change in time when the option
is written (you can think that the letter T represents time). Let t represent the
time when the option is written and T the expiration date. Then
C (T t)
=
t
12.3.5
Rho
12.3.6
Psi
12.3.7
60
S
1 2
1
2
ln
ln
+ r+ T
+ 0.06 0.04 + 0.25 0.75
K
2
65
2
=
d1 =
=
0.25 0.75
T
0.192 2
N (d1 ) = N (0.192 2) = 1 N (0.192 2)
N (0.192 2) = 0.576 2
N (d1 ) = 1 0.576 2 = 0.423 8
N (d1 ) = N (0.192 2) = 0.576 2
113
=e
40
S
1 2
1
2
ln
ln
+ r+ T
+ 0.08 0.03 + 0.2 0.25
K
2
35
2
d1 =
=
=
0.2 0.25
T
1. 510 3
N (d1 ) = 0.934 5
call = eT N (d1 ) = e0.03(0.25) 0.934 5 = 0.927 5
T
Calculate put
N (d
1)
= e
40
1 2
1
S
ln
+ r+ T
+ 0.08 0.03 + 0.22 0.75
ln
K
2
45
2
=
d1 =
=
0.2 0.75
T
0.376 9
N (d1 ) = 0.646 9
put = e0.03(0.75) 0.646 9 = 0.632 5
portf olio = 20 (0.927 5) 35 (0.632 5) = 40. 687 5
Since your write 35 puts, the delta of 35 puts is 35 (0.632 5).
12.3.8
The elasticity is
% change in option price
=
% change in stock price
Suppose the stock price increase by where
Then the option price will change by .
% change in stock price=
S
S
C
=
=
C
S
(12.17)
(12.18)
114
Example 12.3.4. Calculate the elasticity and the volatility of a European call
option and an otherwise identical European put option. The information is:
S = 80,K = 70, = 0.35, r = 5%, = 3%, and T = 0.5.
1
1
80
S
+ r + 2 T
+ 0.05 0.03 + 0.352 0.5
ln
ln
K
2
70
2
d1 =
=
=
0.35 0.5
T
0.703 7
N (d1 ) =
1 0.759 2 = 0.240 8
N (d1 ) = 0.759
2
d2 = d1 T = 0.703 7 0.35 0.5 = 0.456 2
N (d2 ) = 1 0.675 9 = 0.324 1
N (d2 ) = 0.675 9
C = SeT N (d1 )KerT N (d2 ) = 80e0.03(0.5) 0.759 270e0.05(0.5) 0.675 9 =
13. 687
P = SeT N (d1 )+KerT N (d2 ) = 80e0.03(0.5) 0.240 8+70e0.05(0.5)
0.324 1 = 3. 150
call = eT N (d1 ) = e0.03(0.5) 0.759 2 = 0.747 9
put = eT N (d1 ) = e0.03(0.5) 0.240 8 = 0.237 2
call S
0.747 9 80
call =
=
= 4. 371
C
13. 687
put S
0.237 2 80
=
= 6. 024
put =
P
3. 150
call = stock call  = 0.35 (4. 371) = 1. 530
put = stock put  = 0.35 (6. 024) = 2. 108
12.3.9
Seh + Berh
S
S h
eh =
=
e + 1
erh
C
C
C
eh = eh + (1 ) erh
(12.19)
Equation 12.19 holds for any h. Using the Taylors expansion, we have:
1
1 + h + (h)2 + ...
2
1
1
= 1 + ah + (ah)2 + ... + (1 ) 1 + rh + (rh)2 + ...
2
2
115
For the above equation to hold for any h, it seems reasonable to assume that
1 = + (1 )
h = ah + (1 ) rh
1
1
1
2
2
2
(h) + ... =
(ah) + ... + (1 )
(rh) + ...
2
2
2
So we have h = ah + (1 ) rh or
r = ( r)
(12.20)
The Sharp ratio of an asset is the assets risk premium divided by the assets
volatility:
Sharp Ratio =
(12.21)
Sharp Ratiooption =
( r)
r
=
= Sharp Ratiostock
stock
stock
(12.22)
So the Sharp ratio of an option equals the Sharp ratio of the underlying
stock.
12.3.10
12.4
12.4.1
(12.23)
Example 12.4.1.
You buy a European call option that expires in 1 year and hold it for one
day. Calculate your holding profit. Information is:
The stock price is 40 when you buy the option.
116
S
1 2
ln
+ r+ T
K
2
d1 =
T
40
1
ln
+ 0.08 0 + 0.32 1
40
2
=
= 0.416 7
0.3
1
364
40
1
ln
+ 0.08 0 + 0.32
40
2
365
r
= 0.416 1
d1 =
364
0.3
365 r
364
d2 = d1 T = 0.416 1 0.3
= 0.116 5
365
N (d1 ) = 0.661 33
N (d2 ) = 0.546 37
C = 40e0(364/365) 0.661 33 40e0.08(364/365) 0.546 37 = 6. 274
Suppose you buy the option at t = 0 by paying 6. 285 and sell the option
one day later for 6. 274 . Your holding period profit is:
6. 274 6. 285e0.08(1/365) = 0.01 2
You buy a European call option that expires in 1 year and hold it for 6
months. Calculate your holding profit. Information is:
The stock price is 40 when you buy the option.
The stock price is 40 after 6 months.
K = 40
r = 0.08
117
=0
= 30%
At time zero, you buy a 1year European option. Your purchase price is the
call price. As calculated before, the call price is 6. 285
6 months later,
the call is worth:
40
1
ln
+ 0.08 0 + 0.32 0.5
40
2
d1 =
= 0.294 6
0.3 0.5
1
S
+ r + 2 T
ln
K
2
d1 =
T
40
1
2
ln
+ 0.08 0.02 + 0.3 1
40
2
=
= 0.35
0.3
1
option is worth:
42
1
ln
+ 0.08 0.02 + 0.32 0.5
40
2
= 0.477 5
d1 = =
0.3 0.5
118
N (d1 ) = 0.316 5
N (d2 ) = 0.395 4
P = 42e0.02(0.5) 0.316 5 + 40e0.08(0.5) 0.395 4 = 2. 035
Suppose you buy the option at t = 0 by paying 3. 487 and sell the option 6
months later for 2. 035. Your holding period profit is:
2. 035 3. 487e0.08(0.5) = 1. 59
12.4.2
Calendar spread
The textbook has an intimidating diagram (Figure 12.14). Dont worry about
this diagram. Just focus on understanding what a calendar spread is.
A calendar spread (also called time spread or horizontal spread) is an option
strategy that takes advantage of the deteriorating time value of options. A
calendar spread involves selling one option that has a shorter expiring date and
simultaneously buying another option that has a longer expiration date, with
both options on the same stock and having the same strike price.
Suppose that Microsoft is trading for $40 per share. To have a calendar
spread, you can sell a $40strike call on a Microsoft stock with option expiring
in 2 month. Simultaneously, you buy a $40strike call on a Microsoft stock
with option expiring in 3 months. Suppose the price of a $40strike 2month to
expiration call is $2; the price of a $40strike 3month to expiration call is $5.
So your net cost of having a calendar spread at time zero is $3.
Then as time goes by, suppose the stock price doesnt move much and is still
around $40, then the value of your sold call and purchased call both deteriorate
but at a dierent deteriorating speed. The value of the $40strike 2month to
expiration call deteriorates much faster. With each day passing, this option has
less and less value left. If there are only several days left before expiration, the
value of the sold call will be close to zero.
With each day passing, the value of the $40strike 3month to expiration call
also decreases but at a slower speed.
For example, one month later, the sold call has 1 month to expiration and
is worth only $1. The purchased call has 2month to expiration and is worth
$4.5. Now the calendar spread is worth $3.5. You can close out your position by
buying a $40strike 1month to expiration call (price: $1) and sell a $40strike
2month to expiration call (price: $4.5). If you close out your position, youll
get $3.5.
At time zero, you invest $3 to set up a calendar spread. One month later,
you close out your position and get $3.5. Your profit (assuming no transaction
cost) is $0.5.
Time zero: your cost is $3
$40strike call $40strike call
Value
$2
$5
Time to expiration 2 months
3 months
119
12.5
Implied volatility
12.5.1
120
than the observed option price, use a lower trial and try again. Keep doing
this until you find a such the computed option price equals the observed option
price.
First, lets try = 10%
60
1 2
ln
+ 0.06 0.02 + 0.1 0.75
55
2
d1 =
= 1. 3944
0.1 0.75
1 2
60
+ 0.06 0.02 + 0.2 0.75
ln
55
2
d1 =
= 0.762 2
0.2 0.75
1
60
+ 0.06 0.02 + 0.152 0.75
ln
55
2
d1 =
= 0.965 7
0.15 0.75
12.5.2
Volatility skew
12.5.3
121
12.6
Perpetual American options are excluded from the exam syllabus. Please ignore
this chapter.
I included this section for completeness, but you dont need to read it.
12.6.1
Our task here is to derive the price formula for a perpetual American call with
strike price K.
A perpetual American option never expires and the option holder can exercise the option at any time. Its a typical American option where the expiration
date T = +.
The theoretical framework behind the perpetual option formula is the BlackScholes partial dierential equation (called the BlackScholes PDE). The BlackScholes PDE is Derivatives Market Equation 21.11 (page 682):
1
Vt + 2 S 2 VSS + (r ) SVS rV = 0
2
(Textbook 20.11)
(Textbook 13.10)
Please note that Equation 13.10 assumes = 0. In addition,it uses C (instead of V ) to represent the option price.
The BlackScholes PDE is commonly written as:
V (t, St ) 1 2 2 2 V (t, St )
V (t, St )
+ (r ) St
rV (t, St ) = 0 (12.24)
+ St
t
2
St2
St
122
In the above equation, V (t, St ) is the option price at time t where the stock
price is St .
If you are interested in learning how to derive the BlackScholes PDE, refer
to the textbook. For now lets accept Equation 12.24.
V
For a perpetual option, its value doesnt depends on time. Hence
= 0.
t
The BlackScholes PDE becomes an ordinary dierential equation:
1 2 2 d2 V (t, St )
dV (t, St )
+ (r ) St
rV (t, St ) = 0
St
2
dSt2
dSt
(12.25)
d2 V (t, St )
dV (t, St )
So as long as h2 1 = 0, or h = 1, Equation St2
+St
2
dSt
dSt
V (t, St ) = 0 has a solution.
Of course, if Sth is a solution, aSth must also be a solution.
Similarly, we can guess that the solution to Equation 12.25 is in the form of
V (t, St ) = Sth . Some brilliant thinker guessed the following solution:
h
St
H K h
V (t, St ) = (H K)
=
St = aSth
h
H
(H )
Here H the stock price where exercise is optimal (H is a constant). H K
h
St
is the terminal payo at exercise time.
is an indicator telling us how
H
1 2 2
1 2
h + r hr =0
2
2
s
2
1
1 2
r 2 + 2 2 r
(r )
2
2
h=
2
1 r
= 2
2
s
s
r 1
2
2
1 r
2
2
2r
2
2r
2
123
But how can we find H ? Since the perpetual American option can be exercised at any time, the option holder will choose H such that V (t, St ) =
h
St
dV (t, St )
reaches its maximum value. This requires setting
=
(H K)
H
dH
0.
"
h #
dV (t, St )
St
d
(H K)
=
dH
dH
H
h
i
d
1h
h
)
K
(H
)
(H
= Sth
hdH
i
h
h1
h
= St (1 h) (H ) K (h) (H )
=0
(1 h) (H )h K (h) (H )h1 = 0
(1 h) H K (h) = 0
K (h)
h
H =
=
K
1h
h1
h
1
H K =
K K =
K
h1
h1
So V (t, St ) = (H K)
St
H
K
=
h1
h 1 St
h K
Set t = 0. Let S represent the stock price at time zero (i.e. S = S0 ). Then
the option value at time zero is
h
S
K
h1 S
V (0, S) = (H K)
=
H
h1
h K
But how do we choose h since there are two possible value of h? H K =
1
K. To avoid h 1 becoming negative, we choose the bigger h:
h1
s
2
1 r
1 r
2r
+
+ 2
hcall =
2
2
2
2
St
K H h
V (t, St ) = (K H )
=
St = bSth
h
H
(H )
Once again, we get the following equation:
124
1 2
h (h 1) + (r ) h r = 0
2
1
1 2 2
h + r 2 h r = 0
2
s2
2
2r
1 r
1 r
+ 2
h= 2
2
2
dV (t, St )
= 0:
dH
"
h #
h
i
d
dV (t, St )
S
t
h d
h
(K
H
=
(S
=
)
)
)
(H
)
(K
H
t
dH
dH
H
dH
h
i
d
K (H )h (H )1h
= (St )h
hdH
i
h
h1
h
= (St ) K (h) (H )
(1 h) (H )
=0
Set
K (h) (H )
h1
(1 h) (H )
=0
h
H = K
K (h) (1 h) H = 0
h
h
1
K=
K
K H = 1
h1
1h
To avoid 1 h becoming
we choose the smaller h:
s negative,
2
1 r
2r
1 r
+ 2
hput =
2
2
2
2
Summary of the formulas for perpetual American calls and perpetual American puts:
Cperpetual =
(HCall
K)
HCall
hcall
K
=
hcall 1
hcall 1 S
hcall K
hcall
(12.26)
hcall
K
hcall 1
s
2
1 r
2r
1 r
=
+ 2
+
2
2
2
=
HCall
hcall
Pperpetual = K Hput
Hput
hput
Hput
=
K
1 hput
hput
K
hput 1
hput 1 S
hput K
(12.27)
(12.28)
hput
(12.29)
(12.30)
hput =
1 r
2
2
1 r
2
2
125
2
2r
2
1 2
h (h 1) + (r ) h r = 0
2
(12.31)
(12.32)
Example 12.6.1. Calculate the price of a perpetual American call and the price
of an otherwise identical perpetual American put. The information is as follows.
The current stock price is S = 50. The strike price is K = 45. The continuously
compounded riskfree rate is r = 6%. The continuously compounded dividend
yield is 2%. The stock volatility is = 25%.
Solution.
Solve for h.
1 2
h (h 1) + (r ) h r = 0
2
1
0.252 h (h 1) + (0.06 0.02) h 0.06 = 0
2
h2 = 1. 532 7
h1 = 1. 252 7
Use the bigger h for call and the smaller h for put.
Next, calculate the stock price where exercising the option is optimal.
hcall
1. 252 7
HCall
=
K=
45 = 223. 08
hcall 1
1. 252 7 1
hput
1. 532 7
=
Hput
K=
45 = 27. 23
hput 1
1. 532 7 1
hcall
1. 252 7
S
50
= (223. 08 45)
=
Cperpetual = (HCall K)
HCall
223. 08
27. 35
hput
1. 532 7
S
50
Pperpetual = K Hput
=
(45
27.
23
)
= 7.
Hput
27. 23
00
Tip 12.6.1. The CD attached to the textbook Derivatives Markets has a spreadsheet that calculates the price of a perpetual American call and a perpetual American put. The spreadsheet is titled "optbasic2." You can use this spreadsheet to
double check your solution.
12.6.2
Consider the following barrier option. If the stock price first reaches a preset
price H from below, then the payo of $1 is received. This is a special case of a
126
The value at time zero of $1 received when the stock price first reaches H
from below (i.e. the stock first rises to H) is
h1 =
1 r
2
2
S
H
+
h1
(12.33)
1 r
2
2
2r
2
(12.34)
Similarly, the value at time zero of $1 received when the stock price first
reaches H from above (i.e. the stock first falls to H) is
h2 =
1 r
2
2
h1 and h2 satisfies:
S
H
h2
(12.35)
1 r
2
2
2r
2
1 2
h (h 1) + (r ) h r = 0
2
(12.36)
(12.37)
Example 12.6.2.
Calculate the value of a $1 paid if the stock price first reaches $100 and $60
respectively. The information is:
S = 80
r = 6%
= 2%
= 30%
Solution.
calculate the value of a $1 paid if the stock price first reaches $100
H = 100 > S. So we need to calculate the price of $1 payo when the stock
price first rises to H from below
1 2
h (h 1) + (r ) h r = 0
2
1 2
0.3 h (h 1) + (0.06 0.02) h 0.06 = 0
2
h2 = 1. 100 5
h1 = 1. 211 6
Use the bigger h
h1
1. 211 6
80
S
=
= 0.763
H
100
127
calculate the value of a $1 paid if the stock price first reaches $90
H = 60 < S.So we need to calculate the price of $1 payo when the stock
price first falls to H from above.
h1 = 1. 211 6
h2 = 1. 100 5
Use
the
smaller
h
h2 1. 100 5
S
80
=
= 0.729
H
60
Tip 12.6.2. The CD attached to the textbook Derivatives Markets has a spreadsheet that calculates the price of a barrier option. The spreadsheet is titled "optbasic2." You can use this spreadsheet to double check your solution.
128
Chapter 13
Marketmaking and
deltahedging
13.1
Delta hedging
V
Here is the main idea behind delta hedging. Delta of an option is =
.
S
Hence for a small change in stock price, the change of the option value is approximately V1 V0 (S1 S0 ). Suppose you sell one European call option.
If you hold shares of stock, you are immunized against a small change of the
stock price.
If the stock price goes up from S0 to S1 , then the call will be more valuable
to the buyer and your are exposed to more risk. Suppose the value of the call
goes up from V0 to V1 as the stock price goes up from S0 to S1 , then your
liability will increase by V1 V0 . At the same time, the value of your shares
of stock will go up by (S1 S0 ). Because V1 V0 (S1 S0 ), the increase
of your liability will be roughly oset by the increase of your asset.
However, under delta hedging, you are immunized against only a small
change of the stock price (just like immunization by duration matching assets
and liabilities is only good for a small change of interest rate). If a big change
of stock price knocks o your hedging, youll need to rebalance your hedging.
However, in the real world, continuously rebalancing the hedging portfolio
is impossible. Traders can only do discrete rebalancing.
13.2
Make sure you can reproduce the textbook calculation of deltahedging for 2
days. In addition, make sure you can reproduce the textbook table 13.2 and
13.3
Exam problems may ask you to outline hedging transactions or calculate the
hedging profit.
129
130
The major diculty many candidates face is not knowing how to hedge.
They wonder "Should the marketmaker buy stocks? Should he sell stocks?"
To determine how to hedge a risk, use the following ideas:
The goal of hedging is to break even. If a trader makes money
on option, he must lose money on stock; if he loses money on
option, he must make money on stock.
To determine whether a trader should buy stocks or sell stocks,
ask "If the stock price go up (or down), will the trader make
money or lose money on the option?"
If a trader loses money on the option as the stock price goes up,
then the trader needs to initially own (i.e. buy) stocks. This
way, the value of the traders stocks will go up and the trader
will make money on his stocks. He can use this profit to oset
his loss in the option.
If a trader makes money on the option as the stock price goes
up, then the trader needs to initially short sell stocks. This way,
as the stock price goes up, the trader will lose money on the
short sale (because he needs to buy back the stocks at a higher
price). His loss in short sale can oset his profit in option.
Example 13.2.1. The trader sells a call. How can he hedge his risk,?
If the stock price goes up, the call payo is higher and the trader will lose
money. To hedge this risk, the trader should buy stocks. This way, if the stock
price goes up, the trader makes money in the stocks. This profit can be used to
oset the traders loss in the written call.
You can also ask the question "If the stock goes down, will the trader make
money or lose money on the option?" If the stock goes down, the call payo is
lower and the trader will make money. To eat up his profit in the option, the
trader needs to buy stocks. This way, as the stock price goes down, the value
of the traders stocks will go down too and the trader will lose money in his
stocks. This loss will oset the traders profit in the written call.
Example 13.2.2. The trader sells a put. How can he hedge his risk?
If the stock price goes down, the put payo is higher and the trader will
make money. To hedge his risk, the trader should short sell stocks. This way,
if the stock price goes down, the trader can buy back stocks at lower price,
making a profit on stocks. This profit can be used to oset the traders loss in
the written put.
You can also ask the question "If the stock goes up, will the trader make
money or lose money on the option?" If the stock goes up, the put payo is
lower and the trader will make money on the written put. To eat up his profit
in the option, the trader needs to short sell stocks. This way, as the stock price
131
goes up, the trader needs to buy back stocks at a higher price. The trader will
lose money in his stocks. This loss will oset the traders profit in the written
put.
Example 13.2.3. The trader buys a call. How can he hedge his risk?
If a trader buys a call, the most he can lose is his premium and theres no
need to hedge. This is dierent from selling a call, where the call seller has
unlimited loss potential.
However, if a trader really wants to hedge his limited risk, he can short sell
stocks.
Example 13.2.4. The trader buys a put. How can he hedge his risk?
If a trader buys a put, the most he can lose is his premium and theres no
need to hedge. This is dierent from selling a put, where the call seller has a
big loss potential.
However, if a trader really wants to hedge his limited risk, he can buy stocks.
Example 13.2.5.
Reproduce the textbook example of deltahedging for 2 days. Here is the
recap of the information. At time zero the market maker sells 100 European
call options on a stock. The option expires in 91 days.
K = $40
r = 0.08
=0
= 0.3
The stock price at t = 0 is $40
The stock price at t = 1/365 (one day later) is $40.50
The stock price at t = 2/365 (two days later) is $39.25
The marketmaker delta hedges its position daily. Calculate the marketmakers daily marktomarket profit
Solution.
First, lets calculate the call premium and delta at Day 0, Day 1, and Day 2.
I used my Excel spreadsheet to do the following calculation. If you cant fully
match my numbers, its OK.
132
1
St
+ 0.08 + 0.32 T
ln
40
2
d1 =
0.3 T
N (d
1 )
d2 = d1 0.3 T
N (d2 )
C = St N (d1 ) 40e0.08T N (d2 )
= e(T t) N (d1 )
Day 0 (t = 0)
91/365
40
Day 1
90/365
40.50
Day 2
89/365
39.25
0.208048
0.290291
0.077977
0.582404
0.058253
0.523227
2.7804
0.58240
0.614203
0.141322
0.556192
3.0621
0.61420
0.531077
0.070162
0.472032
2.3282
0.53108
Day 2 t = 2/365
T2 = 89/365
S2 = 39.25
C2 = 232.82
2 = 53.108
Beginning of Day 0
Trader #0 goes to work
Day 0 t = 0
T0 = 91/365
S0 = 40
C0 = 278.04
0 = 58.240
Trader #0 goes to work. The brokerage firm (i.e. the employer of Trader
#0) gives Trader #0 C0 = $278. 04. This is what the call is worth today.
Trader #0 needs to hedge the risk of the written call throughout Day 0.
To hedge the risk, Trader #0 buys 0 stocks, costing 0 S0 = 58. 2440 = $
2329. 6. Since Trader #0 gets $278. 04 from the brokerage firm, he needs to
borrow:
0 S0 C0 = 2329. 6 278. 04 = $2051. 56.
The trader can borrow $2051. 56 from a bank or use this own money. Either
way, this amount is borrowed. The borrowed amount earns a risk free interest
rate.
Now Trader #0s portfolio is:
Component
Value
0 = 58.24 stocks 2487. 51
call liability
278. 04
borrowed amount 2051. 56
Net position
0
End of Day 0 (or Beginning of Day 1)
133
134
0.5
On Day 0, the trader owns 0 = 58.240 stocks to hedge the call liability
C0 = $278. 04. The traders net asset is
M V (0) = 0 S0 C0 = 58. 24 40 278. 04 = $2051. 56
In the end of Day 0 (or the beginning of Day 1) before the trader rebalances
his portfolio, the traders asset is:
M V BR (1) = 0 S1 C1 = 58. 24 40.5 100 3.0621 = 2052. 51
BR stands for before rebalancing.
The traders profit at the end of Day 0 is:
M V BR (1) M V (0) e0.081/365 = (0 S1 C1 ) (0 S0 C0 ) erh = 2052.
51 2051. 56e0.081/365 = 0.50
Please note that Trader #0 doesnt need to rebalance the portfolio. The
portfolio is rebalanced by the next trader.
Beginning of Day 1
Trader #1 goes to work
Day 1 t = 1/365
T1 = 90/365
S1 = 40.50
C1 = 306.21
1 = 61.420
Trader #1 goes to work. He starts from a clean slate. The portfolio is
automatically rebalanced since Trader #1 starts from scratch.
The brokerage firm gives Trader #1 C1 = $306.21.
Trader #1 needs to hedge the risk of the written call throughout Day 1.
To hedge the risk, Trader #1 buys 1 = 61.420 stocks, costing 1 S1 =
61.42 40.5 = $2487. 51. Since Trader #1 gets $306.21 from the brokerage firm,
he needs to borrow:
1 S1 C1 = 2487. 51 306.21 = 2181. 3
The trader can borrow $2181. 3 from a bank or use this own money. Either
way, this amount is borrowed. The borrowed amount earns a risk free interest
rate.
Now Trader #1s portfolio is:
135
value
2487. 51
306.21
2181. 3
0
One question arises, "What if Trader #1 doesnt start from a clean slate?"
Next, well answer this question.
Instead of starting from scratch, Trader #1 can start o with Trader #0s
portfolio. At the end of Day 0, Trader #0 has
0 = 58.24 stocks
a borrowed amount (0 S0 C0 ) erh = 2051. 56e0.081/365 = 2052. 01
0.5 profit
At the end of Day 0 or the beginning of Day 1, 1 = 61.420. So Trader #1
buys additional shares:
1 0 = 61.42 58.24 = 3. 18
The cost of these additional shares is (1 0 ) S1 = 3. 18 40.5 = 128. 79
Since these additional shares are bought at the current market price, Trader
#1 can sell these shares at the same price he bought them. This doesnt aect
the marktomarket profit.
Trader #1 can borrow 128. 79 to pay for the purchase of 3. 18 stocks.
Now Trader #1 has a total of 1 = 61.42 shares worth 1 S1 = 61.4240.5 =
2487. 51
His liability is now C1 = 306.21
The borrowed amount is now 2052. 01 + 128. 79 + 0.5 = 2181. 3. The 0.5 (the
profit made by Trader #0) is the amount of money Trader #1 borrows from
Trader #0.
Now Trader #1s portfolio is:
component
value
1 = 61.42 stocks $2487. 51
call liability
306.21
borrowed amount 2181. 3
Net position
0
The portfolio is the same as the portfolio if Trader #1 starts from scratch.
Calculation is simpler and cleaner if we start from scratch.
End of Day 1 (or Beginning of Day 2)
Mark to market without rebalancing the portfolio
Day 2 t = 2/365
T2 = 89/365
S2 = 39.25
C2 = 232.82
136
Method 1
To cancel out his position, the trader can
buy a call from the market paying C2 = 232.82. The payo of this call
will exactly oset the payo of the call sold by the brokerage firm to the
customer. These two call have the common expiration date T1 = 89/365
and the same payo. They will cancel each other out.
sell out 1 = 61.420 stocks for 1 S2 = 61.420 39.25 = 2410. 735
pay o the loan. The payment is (1 S1 C1 ) erh = 2181. 3e0.081/365 =
2181. 778
The traders profit at the end of Day 1 is
C2 + 1 S2 (1 S1 C1 ) erh = 232.82 + 2410. 735 2181. 778 = 3. 863
Trader #1 hands in 3. 863 profit to his employer and goes home.
Method 2
We consider the change between the beginning of Day 1 and
the end of Day 1 (or the beginning of Day 2).
In the beginning of Day 1, the traders stock is worth 1 S1 ; In the end of
Day 1 (or the beginning of Day 2), the traders stock is worth 1 S2 . The
value of the traders stocks goes up by 1 S2 1 S1
In the beginning of Day 1, the call is worth C1 ; In the end of Day 1 (or the
beginning of Day 2), the call is worth to C2 . The call value is the traders
liability. Now the traders liability increases by C2 C1
In the beginning of Day 1, the borrowed amount is 1 S1 C1 . In the
end of Day 1 (or the beginning of Day 2), this borrowed
amount grows
to (1 S1 C1 ) erh . The increase is (1 S1 C1 ) erh 1 . This is the
interest paid on the amount borrowed. No matter the trader borrows
money from a bank or uses his own money, the borrowed money needs to
earn a risk free interest rate.
The traders profit at the end of Day 1:
(1 S2 1 S1 ) (C2 C1 ) (1 S1 C1 ) erh 1
= C2 + 1 S2 (1 S1 C1 ) erh = 3. 863
Method 3
In the beginning of Day 1, the traders net asset is
M V (1) = 1 S1 C1 = 61.420 40.5 306.21 = 2181. 3
In the end of Day 1 before the trader rebalances his portfolio, the traders
asset is:
M V BR (2) = 1 S2 C2 = 61.420 39.25 232.82 = 2177. 915
The traders profit at the end of Day 1 is:
M V BR (2) M V (1) erh = 2177. 915 2181. 3e0.081/365 = 3. 863
137
= (1 S2 1 S1 ) (C2 C1 ) (1 S1 C1 ) erh 1
= C2 + 1 S2 (1 S1 C1 ) erh
Method 3 is often faster. Under this method,
Profit during a day=Asset at the end of Day before rebalancing the portfolio
 Future value of the asset at the beginning of the day
Asset at the end of Day before rebalancing the portfolio = delta at the
beginning of the day stock price at the end of the day
Profit at the end of Day t = M V BR (t + 1)M V (t) erh = (t St+1 Ct+1 )
(t St Ct ) erh
Now this completes the textbook example on page 417 to 418.
Next, lets do additional calculations and calculate the profit at the end of
Day 2, 3, and 4 (or the profit at the beginning of Day 3,4,5)
Day
Time t
Expiry T
St
Ct
t
Profit end of Day
Day 2
t = 2/365
T2 = 89/365
S2 = 39.25
C2 = 232.82
2 = 53.108
0.40
Day 3
t = 3/365
T3 = 88/365
S3 = 38.75
C3 = 205. 46
3 = 49. 564
4. 0
Day 4
t = 4/365
T4 = 87/365
S4 = 40
C4 = 271.04
4 = 58.06
1. 32
Day 5
t = 5/365
T5 = 86/365
S5 = 40
C5 = 269.27
5 = 58.01
138
13.3
M V (t) erh 1
= M V BR (t + 1) M V (t) +
{z
}

{z
}

capital gain at end of Day t
Define M V (t) erh 1 as the interest earned at the end of Day t. If the
trader invests
money
at the beginning of Day t, then M V (t) is positive and
M V (t) erh 1 is negative. The negative interest earned is just the interest
expense incurred by the trader.
Define M V (t) as the investment made at the beginning of Day t. If M V (t)
is negative, then it means that the trader receives money.
Example 13.3.1.
Reproduce the textbook Table 13.2
We already reproduced the daily profit in Table 13.2. We just need to
reproduce the investment, interest, and the capital gain.
139
Day
Time t
Expiry T
St
Ct
t
Investment (beginning of the day)
interest earned during the day
Capital gain (end of the day)
Profit (end of the day)
0
t=0
T0 = 91/365
S0 = 40
C0 = 278.04
0 = 58.2404
2051. 56
0.45
0.95
0.50
1
t = 1/365
T1 = 90/365
S1 = 40.5
C1 = 306.21
1 = 61.42
2051. 56
3. 385
0.478
3. 863
2
t = 2/365
T2 = 89/365
S2 = 39.25
C2 = 232.82
2 = 53.108
1851. 669
0.41
0.81
0.40
Day 0
We already know:
M V (0) = 0 S0 C0 = 58. 24 40 278. 04 = $2051. 56
M V BR (1) = 0 S1 C1 = 58. 24 40.5 100 3.0621 = 2052. 51
The traders profit at the end of Day 0 is:
M V BR (1) M V (0) erh = 2052. 51 2051. 56e0.081/365 = 0.50
To find the capital gain and the interest earned at the end of Day 0, we just
need to break down the profit M V BR (1) M V (0) erh into two parts:
M V BR (1) M V (0) erh
= M V BR (1) M V (0) M V (0) erh +
M V (0)
rh
BR
= MV
(1) M V (0) + M V (0) e 1
{z
}

{z
}

capital gain
interest earned
The interest
credited
at the endof Day 0:
140
To find the capital gain and the interest earned at the end of Day 0, we just
need to break down the profit M V BR (1) M V (0) erh into two parts:
interest earned
Capital gain at the end of Day 1: M V BR (2) M V (1) = 2177. 915 2181.
3 = 3. 385
13.4
5
t = 5/365
T5 = 86/365
S5 = 40
C5 = 269.27
5 = 58.01
The stock price of Table 13.3 follows the binomial tree with = 0.3.
On Day 0, the stock price is S0 = 40
On Day 1 thestock moves up 1
S1 = S0 erh+ h = 40e0.08/365+0.3 1/365 = 40.642
Day 2 the stock
moves down 1
rh h
S2 = S1 e
= 40.642e0.08/3650.3 1/365 = 40. 018
Day 3 the stock
moves down 1
rh h
S4 = S3 e
= 39. 403e0.08/3650.3 1/365 = 38. 797
Day 5 the stock
moves up 1
141
interest, the capital gain, and the daily profit on any other day is the interest,
the capital gain, and the daily profit at the end of the previous day or the
beginning of that day.
The author of the textbook uses Table 13.3 to show us that if the stock price
moves up or down 1 daily, then the traders profit is zero.
13.5
13.5.1
DeltaGammaTheta approximation
(Textbook 13.6)
V (St , T t) 2
1 V (St , T t) 2
+
h
+
2
S 2
2
t2
However,
V (St , T t)
= t
S
2 V (St , T t)
= t
S 2
1 2 V (St , T t) 2
h since its close to zero. However,
2
t2
1 2 V (St , T t) 2
is not close to zero. The reason that
h
2
t2
We decide to ignore
1 2 V (St , T t)
2
S 2
V (St , T t)
=
t
142
1 2 V (St , T t) 2
is close to zero but
is not close to zero will be explained
2
S 2
in Derivatives Markets Chapter 20 when we derive Itos Lemma. For now just
accept it.
Now we have:
1
V (St+h , T t h) V (St , T t) + t + h + t
2
13.5.2
Suppose a trader sets up a hedging portfolio at time t. The traders profit after
a short interval h (i.e. at time t + h) is:
P rof it (t + h)
= M V BR (t + h) M V (t) erh
= (t St+h Ct+h ) (t St Ct ) erh
1
P rof it (t + h) = t (St+h St ) t + h + t 2
2
Since St+h = St + , we have:
P rof it (t+ h)
1 2
= t t + h + t
rh (t St Ct )
2
1 2
rh (t St Ct )
= h + t
2
1 2
t + h rh [t St Ct ]
P rof it (t + h)
2
However, St+h = St erh+
rh (t St Ct )
(Textbook 13.7)
1
2
= 1+ rh + h + rh + h +
2
2
1
rh + h and higher order terms all approach zero.
2
h.
As h 0,
143
1
2
= St 1 + rh + h +
rh + h + St 1 + h
2
= St+h St St h
2 St2 2 h
Plug the above equation in Textbook Equation 13.7, we get:
St+h = St erh+
P rof it (t + h)
1 2 2
St ht + h rh [t St Ct ]
2
(Textbook 13.9)
From the textbook Table 13.3, we know that if the stock price moves up or
down
by 1 , the traders profit is zero. So
1 2 2
S ht + h rh [t St Ct ] = 0
2 t
This gives us the BlackScholes PDE:
1 2 2
S t + r [t St Ct ] = 0
2 t
(Textbook 13.10)
144
Chapter 14
Exotic options: I
Any option that is not a plain vanilla call or put is called an exotic option.
There are usually no markets in these options and they are purely bought OTC
(overthecounter). They are much less liquid than standard options. They often
have discontinuous payos and can have huge deltas near expiration which make
them dicult to hedge.
Before studying this chapter, make sure you understand the learning objective.
SOAs learning outcome for this chapter:
Explain the cash flow characteristics of the following exotic options: Asian,
barrier, compound, gap and exchange
If you want to cut corners, you can skip the pricing formula for exotic options
because calculating the exotic option price is out of the scope of the learning
outcome or learning objective.
14.1
14.1.1
Characteristics
Asian options (also called average options) have payos that are based on the
average price of the underlying asset over the life of the option. The average
price can be the average stock price or the average strike price. The average
can be arithmetic or geometric.
The unique characteristic of an average price option is that the underlying
asset prices are averaged over some predefined time interval.
Averaging dampens the volatility and therefore average price options are
less expensive than standard options
145
146
Average price options are pathdependent, meaning that the value of the
option at expiration depends on the path by which the stock arrives at its
final price. The price path followed by the underlying asset is crucial to
the pricing of the option.
Average price options are useful in situations where the trader/hedger
is concerned only about the average price of a commodity which they
regularly purchase.
14.1.2
Examples
A 9month European average price contract calls for a payo equal to the
dierence between the average price of a barrel of crude oil and a fixed
exercise price of USD18. The averaging period is the last two months of
the contract. The impact of this contract relative to a standard option
contract is that the volatility is dampened by the averaging of the crude oil
price, and therefore the option price is lower. The holder gains protection
from potential price manipulation or sudden price spikes.
A Canadian exporting firm doing business in the U.S. is exposed to Can$/US$
foreign exchange risk every week. For budgeting purposes the treasurer
must pick some average exchange rate in which to quote Can$ cash flows
(derived from US$ revenue) for the current quarter. Suppose the treasurer
chooses an average FX rate of Can$1.29/US$1.00. If the US$ strengthens, the cash flows will be greater than estimated, but if it weakens, the
companys Can$ cash flows are decreased.
Arithmetic average
We record the stock price every h periods from time 0 to time T . There are
N = T /h periods. The arithmetic average is:
n
1 X
A (T ) =
Sih
N i=1
14.1.3
Geometric average
14.1.4
147
Payo at maturity T
14.2
price call
price put
strike call
strike put
Payo
max [0, AV G (T ) K]
max [0, K AV G (T )]
max [0, ST AV G (T )]
max [0, AV G (T ) ST ]
Barrier option
Barrier options are similar to standard options except that they are extinguished
or activated when the underlying asset price reaches a predetermined barrier or
boundary price. As with average options, a monitoring frequency is defined as
part of the option which specifies how often the price is checked for breach of
the barrier. The frequency is normally continuous but could be hourly, daily,
etc.
14.2.1
Knockin option
"In" option starts its life worthless and becomes active only if the barrier
price is reached
Technically, this type of contract is not an option until the barrier price
is reached. So if the barrier price is never reached it is as if the contract
never existed.
downandin (spot price starts above the barrier level and has to move
down for the option to become activated.)
upandin (spot price starts below the barrier level and has to move up
for the option to become activated.)
14.2.2
Knockout option
"Out" option starts its life active and becomes null and void if the barrier
price is reached
The option will expire worthless if the asset price exceeds the barrier price
downandout (spot price starts above the barrier level and has to move
down for the option to become null and void)
upandout (spot price starts below the barrier level and has to move up
for the option to be knocked out)
148
14.2.3
Rebate option
14.2.4
Barrier parity
(14.1)
Example 14.2.1.
Explain why
upandin option + upandout option = Ordinary option.
T =option expiration time
tH =time when the stock price first reaches the barrier, where the barrier
price is greater than the current stock price
During the time interval [0, tH )
the upandout option is alive
the upandin option is dead
During the time interval [tH , T ]
the upandout option is dead (because the barrier is reached)
the upandin option is alive (because the barrier is reached)
If we have a downandin option and a downandout option with two options
on the same stock, having the same barrier price and the same strike price, then
at any moment in the interval [0, T ], well always have exactly one ordinary
option alive. Hence downandin option + downandout option = Ordinary
option.
Similarly, downandin option + downandout option = Ordinary option.
14.2.5
Examples
149
14.3
Compound option
Definition
With a compound option one has the right to buy an ordinary option at
a later date
Compound options are more expensive to purchase than the underlying
option, as the purchaser has received a price guarantee and eectively
extended the life of the option
Make sure you understand the textbook Figure 14.2.
For a compound call (i.e. call on call) to be valuable, the following two
conditions need to be met:
St1 > S
St1 > K
Compound option parity
The key formula is DM 14.12:
CallOnCall P utOnCall + xert1 = BSCall
(DM 14.12)
This formula looks scary but is actually easy to remember. We know the
putcall parity C + KerT = P + S0 . If we treat the standard BSCall as the
underlying asset, then applying the standard putcall parity, we get:
CallOnCall + xert1 = P utOnCall + BSCall
This is DM 14.12.
Similarly, we have:
CallOnP ut + xert1 = P utOnP ut + BSP ut
150
(DM 14.13)
= St1 + D K + max PE (St1 , T t1 ) D K 1 er(T t) , 0
i
CA (St1 , T t1 ) = St1 +DK+max PE (St1 , T t1 ) D K 1 er(T t) , 0
(DM 14.14)
max PE (St1 , T t1 ) D K 1 er(T t) , 0 is payo of a call on put
151
DM Example 14.2 shows you how to use DM 14.14 to calculate the price of
an American call option with a single dividend. When reading this example,
please note two things:
(1) DM Example 14.2 has errors. Make sure you download the errata. The
DM textbook errata can be found at the SOA website.
(2) The call on put price is calculated using the Excel spreadsheet (so dont
worry about how to manually calculate the call on put).
14.4
Gap option
14.4.1
Definition
An option in which one strike price K1 determines the size of the payo
and another strike price K2 determines whether or not the payo is made.
Example. A call option pays ST 5 if ST > 8 and pays zero otherwise.
The strike price K1 = 5 determines the size of the payo but a dierent
strike price K2 = 8 determines whether or not the payo is made. K2 is
called the payment trigger.
Example. A put option pays 8 ST if ST > 5 and pays zero otherwise.
The strike price K1 = 8 determines the size of the payo but a dierent
strike price K2 = 8 determines whether or not the payo is made.
As the footnote in Page 457 of the textbook indicates, a gap call or put
option is really not a option; once the payment trigger is satisfied, the
owner of a gap option MUST exercise the option. Hence the premium of
a gap option can be negative.
14.4.2
Pricing formula
To find the price of a gap option, you can modify a standard options formula
by changing d1 :
C (K1 , K2 ) = SeT N (d1 ) K1 erT N (d2 )
rT
T
N (d
N (d1 )
P (K1 , K2 ) = K
2 ) Se
1 e
1 2
S
+ r+ T
ln
K2
2
d1 =
T
d2 = d1 T
14.4.3
152
1 2
40
S
1
2
+ r+ T
ln
ln
+ 0.06 0.02 + 0.3 0.5
K2
2
50
2
=
d1 =
=
0.3 0.5
T
0.851 6
14.5
Exchange option
Allows the holder of the option to exchange one asset for another
Pricing
p formula. Use the standard BlackScholes formula except changing
to 2S + 2K 2 S K . Also, make sure you know which is the stock
asset and which is the strike asset. If you give up Asset 2 and receive
Asset 1, Asset 1 is the stock and Asset 2 is the strike asset.
153
This is slightly dierent from the textbook call price 6.6133 due to rounding.
You can verify the call price using the Excel worksheet "Exchange."
Inputs:
Underlying Asset
Price
95.92
Volatility
20.300%
Dividend Yield
0.750%
Strike Asset
Price
95.92
Volatility
22.270%
Dividend Yield
1.170%
Other
Correlation
0.6869
Time to Expiration (years)
Output:
Exchange Option
BlackScholes
Call
Put
Price
6.6144
6.2154
So the exchange call price is 6.6144; the exchange put price is 6.2154.
154
Chapter 18
Lognormal distribution
This chapter is an easy read. Im going to highlight the major points.
18.1
Normal distribution
The normal distribution has the following pdf (probability density function)
(DM 18.1):
"
2 #
1
1 x
(x; , ) = exp
2
2
In the standard normal
distribution,
= 0 and = 1. Its pdf is:
1
1 2
(x; 0, 1) = exp x
2
2
Its cdf (cumulative probability density function) is DM18.2:
Ra
Ra
1
1
P (z 6 a) = N (a) = (x; 0, 1) dx = exp x2 dx
2
2
A frequently used formula is DM 18.3:
N (a) = 1 N (a)
How to convert a normal
random variable to the standard normal random
variable. If x N , 2 , then we can transform x into a standard normal
random variable using DM 18.4:
x
z=
1
1
x is normal
(linear combination
ofnormal random variables are also normal). Next,
x
1
E (z) = E
= [E (x) ] = 0
155
156
V ar (z) = V ar
So z N (0, 1)
1
1
1
V ar (x ) = 2 V ar (x) = 2 2 = 1
2
normal random
2
N 1 , 21 and x2
N
,
then
2
2
Next, the textbook briefly explains the central limit theorem, a concept you
most likely already know.
18.2
Lognormal distribution
2
2
If x N m, v 2 , then E (ex ) = em+0.5v and V ar (ex ) = e2m+v ev 1 .
You dont need to know how to derive these formulas. Just memorize them.
18.3
Define:
, the expected (annualized) continuously compounded return earned by
the stock
, the (annualized) continuously compounded dividend yield
, the stocks volatility
z, the standard normal random variable
Then the (not annualized) continuously
compounded
return earned during
t (DM 18.18):
St
N 0.5 2 t, 2 t
ln
S0
random variable with mean
Using DM2 18.7, we can rewrite the normal
157
x = 0.52 t + tz
St
ln
= x = 0.5 2 t + tz
S0
St = S0 ex = S0 exp 0.5 2 t + tz
Using DM 18.13,
we
get DM 18.22:
18.4
ln K ln S0 0.5 2 t
P (St < K) = N
= N d 2
t
S0
ln
+ 0.5 2 t
ln K ln S0 0.5 2 t
K
=
where
=
t
t
d2
ln K ln S0 + 0.52 t
In the above formula, if we set = r, well get the risk neutral probability
of St < K:
"
ln K ln S0 + r 0.5 2 t
P (St < K) = N
= N (d2 )
t
Next, lets calculate the real world probability of P (St > K). Please note
that P (St = K) = 0. This is because St is a continuous random variable; the
probability for a continuous random variable to take on a specific value is zero.
So P (St > K) = 1 P (St < K) P (St = K) = 1 P (St < K). The real
world probability of St > K is:
S0
2
ln
t
+
0.5
= N d2 = N K
P (St > K) = 1N d2 = 1 1 N d2
158
If we set = r, well get the risk neutral probability of P (St > K):
S0
ln
+ r 0.52 t
18.4.1
1p
1+p
1
1
N
<z<N
.
2
2
For example, if p = 0.95, then the confidence interval is:
1 0.95
1 + 0.95
< z < N 1
, which is N 1 (0.025) < z <
2
2
N 1 (0.975) , which is 1.96 < z < 1.96.
N 1
normal random
1 (1 p)
1 + (1 p)
1
1
1 p
is N
< z < N
, which is N
< z <
2
2
2
p
N 1 1
.
2
For example, if p = 5%, then the 1 5% = 95% confidence interval is:
159
0.05
0.05
< z < N 1 1
, which is N 1 (0.025) < z < N 1 (0.975),
2
2
which is 1.96 < z < 1.96.
N 1
p
1
+
p
+ N 1
< x < + N 1
.
2
2
For example, the 95% confidence interval of x N 1, 22 is: 1+2 (1.96) <
x < 1 + 2 (1.96), which is 2. 92 < x < 4. 92.
p
+ N 1
< x < + N 1 1
2
2
1p
You dont need to memorize messy formulas such as + N 1
<
2
1+p
p
x < + N 1
or + N 1
< x < + N 1 1
. The
2
2
2
The 99% confidence interval for z is 2.576 < z < 2.576; for x N 1, 22
is 1 2.576 2 < x < 1 + 2.576 2, which is 4. 152 < x < 6. 152.
Now lets walk through DM example 18.6. Here we need to find the 95%
confidence interval for the stock price St . The inputs are:
S0 = 100
t=2
160
= 0.1
= 0.3
=0
St
We know that ln
N a 0.5 2 t, 2 t .
S0
S2
< 0.11 + 1.96 0.18,
Answer: 0.11 1.96 0.18 < ln
100
S2
or 0.721 56 < ln
< 0.941 56.
100
Step 2
S2
Take exponentiation of 0.721 56 < ln
< 0.941 56
100
S2
exp (0.721 56) < exp ln
< exp (0.941 56)
100
Step 3
S2
exp ln
=
100
S2
< 2. 563 98
48.60 < S2 < 256.40
100
So the 95% confidence interval for S2 is (48.60, 256.40) .
0.485 994 <
Now lets walk through Row 1 (i.e. 1 day horizon) in DM Table 18.1. The
inputs are:
S0 = 50
t = 1/365 (i.e. 1 day)
= 0.15
=0
= 0.3
161
St
N a 0.5 2 t, 2 t
S0
2
S1/365
0.15 0 0.5 0.32 0.32
4 0.3
ln
N
,
= N 2. 876 71 10 ,
50
365
365
365
ln
Whats the confidence interval for the standard normal random variable z?
Answer: 1 < z < 1.
Whats the corresponding confidence interval for ln
Answer:
S1/365
?
50
S1/365
0.32
< ln
< 2. 876 71 104 + 1
2. 876 71 10 1
365
50
S1/365
1. 541 505 102 < ln
< 1. 599 04 102
50
2
2
50e1. 541 50510 < S1/365 < 50e1. 599 0410
4
0.32
365
162
18.4.2
RK
()t
St g (St ; S0 ) dSt = E (St ) N d1 = S0 e
N d 1
0
S0
+ + 0.5 2 t
K
=
t
g (St ; S0 ) is the probability density function of St . Here St ; S0 indicates that
the initial stock price for St is S0 .
ln
d1
RK
S0 e()t N d1
St g (St ; S0 ) dSt
P (St < K)
N d2
Once you memorize DM 18.27, you can also derive DM 18.29:
R
K
R
0
RK
0
St g (St ; S0 ) dSt =
R
0
St g (St ; S0 ) dSt
RK
0
St g (St ; S0 ) dSt
= K St g (St ; S0 ) dSt = S0 e
1 N d1
= S0 e()t 1 1 N d1
= S0 e()t N d1
R
()t
18.4.3
R
K
()t
N d1
1 N d 1
S0 e
1 N d2
N d2
()t
St g (St ; S0 ) dSt
=
P (St > K)
S0 e
BlackScholes formula
This section derives the BlackScholes formula with simple math. The BlackScholes formula was originally derived using stochastic calculus and partial differential equation. Many years after the BlackScholes formula was published,
someone came up with this simple proof (hindsight is always 2020).
S0 e(r)t N (d1 )
N (d2 )
E (KSt > K) = K
=
E (KSt > K) P (St > K) = KN (d2 )
=
18.5
hz
ln
St
= 0.5 2 h + hz
Sth
h
i
St
=
E ln
= E 0.52 h + hz = 0.5 2 h+
Sth
2
h
hE (z) = 0.5
h
i
St
=
V ar ln
= V ar 0.5 2 h + hz
Sth
V ar 0.5 2 h + hz = h V ar (z) = 2 h
Now lets go through DM Table 18.2.
164
week
1
2
3
4
5
6
7
sum
100
105.04
105.76
108.93
102.5
104.8
104.13
0
St
ln
St1
0.049171
0.006831
0.029533
0.060843
0.022191
0.006414
0.040470
St
ln
St1
0.0024178
0.0000467
0.0008722
0.0037018
0.0004924
0.0000411
0.0075721
Wehave:
St
0.040470
E ln
= 0.5 2 h =
= 0.006 745
Sth
6
0.006 745
= = + 0.5 2 +
h
In the above equations,
is the stocks expected (annual) continuously compounded return
is the stocks (annual) continuously compounded dividend yield ( = 0 in
this problem)
is the stocks (annual) volatility
h = 1/52 (52 weeks in one year)
2
P
St
ln
St
n
St1
1 P ln St
=
V ar ln
= 2h =
Sth
n 1 n
St1
n
6
5
1
0.0075721 0.006 7452
6
=
=
0.388 7
= 0.001 459 83
0.001 459 83
h
0.006 745
0.001 459 83 0.006 745
= + 0.5 2 +
= 0 + 0.5
+
=
h
1/52
1/52
2 =
1P
1P
St
St
=
=
Please note that E ln
ln
(ln St ln Sth ) =
Sth
n
Sth
n
1 Sn
1
= 0.5 2 h
(ln Sn ln S0 ) = ln
n
n S0
165
In this problem,
we could have estimated
St
1 104.13
1 Sn
E ln
= ln
= 0.52 h = ln
= 0.006745
Sth
n S0
6
100
In addition, we have:
1
1
Sn
Sn
= ln
0.5 2 =
ln
nh S0
T
S0
where T = nh is the length of the observation.
We see that when we estimate 0.5 2 , only the following 3 factors
matter:
the length of the observation T
the stocks starting price S0
the stocks ending price Sn
The stock prices between S0 and Sn are irrelevant. And for a given T ,
increasing the number of observations n doesnt aect our estimate of
0.5 2 (because as n goes up, h goes down, and T = nh is a constant). Frequent
observations dont improve our estimate of 0.5 2 . To improve our
estimate of 0.5 2 , we need to increase T .
When we estimate , the inbetween stock prices do matter and more frequent observations do improve our estimate.
18.6
The lognormal stock price model assumes that the stock returns are normally
distributed. Are stock returns really normally distributed?
18.6.1
Histogram
One method to assess whether stock returns are normally distributed is to plot
the continuously compounded returns as a histogram. Look at DM Figure 18.4.
The histograms in Figure 18.4 dont appear normal. The textbook oers
two explanations:
Stock prices can jump discretely from time to time
Stock returns are normally distributed, but the variance of the return
changes over time.
Thats all your need to know about this section.
166
18.6.2
...
Next, plot the range "A2:B518" as xy diagram in Excel. The plot you got
should look like the 2nd diagram in DM Figure 18.2 (see DM page 589). You
can see that from diagram from z = 2.58 to z = 2.58 is roughly a straight line.
The point of this Excel experiment: if you plot samples from a normal
distribution in a normal probability plot, youll get roughly a straightline. If
you plot samples from an unknown distribution in a normal probability plot and
get roughly a straight line, then the samples are from approximately normal
distribution.
Steps on how to construct a normal probability plot:
1. Sort the samples from smallest to biggest. Number the samples as i =
1, 2, 3, ..., n. So you samples are x1 , x2 , ..., xn .
B518 =
167
i 0.5
.Here
n
1
0.5 is the continuity adjustment. Calculate the corresponding zi = N (yi ).
xi
x1
x2
x3
...
...
xn
yi = P (x < xi ) =
0.5
y1 =
n
1.5
y2 =
n
2.5
y3 =
n
...
n 0.5
yn =
n
i 0.5
n
zi = N 1 (yi )
0.5
z1 = N 1
n
1.5
z2 = N 1
n
2.5
z3 = N 1
n
...
n 0.5
1
zn = N
n
18.7
xi
Sample problems
Problem 1
168
= 0.12
= 0.06
= 0.25
T =2
Construct the 95% confidence interval of ST .
Solution
First we find the 95% confidence interval for the standard normal random
variable z. We need to b such that P (b < z < b) = 0.95.
P (b < z < b) = N (b) N (b) = N (b) [1 N (b)] = 2N (b) 1 = 0.95
1 + 0.95
N (b) =
= 0.975
b = N 1 (0.975) = 1. 96
2
So the 95% confidence interval for z is (1.96, 1.96).
ln
ST
N a 0.5 2 T, 2 T
S0
169
Using the procedure described in the McDonald text, determine the upper
bound of the 90% confidence interval for the price of the stock in 6 months.
Solution
90% confidence interval for a standard normal random variable is (1.645, 1.645).
ST
ln
N a 0.5 2 T, 2 T
S0
ST
is 0.044 375 1.645 0.352 0.5, 0.044 375 + 1.645 0.352 0.5 .
S0
So the upper boundfor ST is:
2
0.25e0.044 375+1.645 0.35 0.5 = 0.392 7
Problem 3
A European call option and a European put option are written on the same
stock. You are given:
S0 = 100
K = 105
= 0.08
=0
= 0.3
T = 0.5
Calculate the probability that the call option will be exercised.
Calculate the probability that the put option will be exercised.
Solution
just need to use
The call will be exercised is S
T > K. We
DM 18.24.
S
0
ln
+ 0.5 2 t
P (ST > K) = N d2 = N K
100
ln
+ 0.08 0 0.5 0.32 0.5
= N 105
0.3 0.5
170
Alternatively,
P (ST < K) = 1 P (ST > K) = 0.558 6
Problem 4
S0 = 50
= 0.1
=0
= 0.4
Calculate the conditional expected value of the stock at T = 3 given ST > 75.
Calculate the conditional expected value of the stock at T = 3 given ST < 75.
Solution
Need to use DM 18.28 and 18.29.
S0 e()t N d1
N d 2
d1
ln
=
S0 e()t N d1
E (St St > K) =
N d2
S0
50
ln
+ + 0.5 2 t
+ 0.1 0 + 0.5 0.42 3
K
75
= 0.194 2
=
t
0.4 3
S0
50
ln
+ 0.5 2 t
+ 0.1 0 0.5 0.42 3
K
75
=
= 0.498 6
=
t
0.4 3
50e(0.10)3 N (0.194 2)
50e(0.10)3 0.423 0
=
= 41. 32
E (St St < K) =
N (0.498 6)
0.6910
d2
03
ln
50e(0.10)3 N (0.194 2)
50e(0.10)3 (1 0.423 0)
==
= 126.
N (0.498 6)
1 0.6910
Problem 5
Stock prices follow lognormal distribution. You are given:
S0 = 100
= 0.08
=0
= 0.3
T =5
171
ST
ln
N a 0.5 2 T, 2 T
S0
75 percentile of ST is:
100ex = 100e0.627 47 = 187. 29
Median is 50 percentile, which corresponds to z = 0 and x = 0.175.
Median of ST is: 100e0.175 = 119. 12
Problem 6 (Spring 2007 C #34)
The price of a stock in seven consecutive months is:
Month Price
1
54
2
56
3
48
4
55
5
60
6
58
7
62
Based on the procedure described in the McDonald text, calculate the annualized expected return of the stock.
(A) Less than 0.28
(B) At least 0.28, but less than 0.29
(C) At least 0.29, but less than 0.30
(D) At least 0.30, but less than 0.31
(E) At least 0.31
Solution
172
price
1
2
3
4
5
6
7
sum
54
56
48
55
60
58
62
St
St1
x2
0.036368
0.154151
0.136132
0.087011
0.033902
0.066691
0.138150
0.001323
0.023762
0.018532
0.007571
0.001149
0.004448
0.056785
x = ln
0.138150
St
= 0.5 2 h =
E ln
= 0.023 025
Sth
6
0.023 025
= = + 0.5 2 +
h
In the above equations,
is the stocks expected (annual) continuously compounded return
is the stocks (annual) continuously compounded dividend yield ( = 0 in
this problem)
is the stocks (annual) volatility
h = 1/12 (12 months in one year)
2
P
St
ln
St
n
St1
1 P ln St
=
V ar ln
= 2h =
Sth
n1 n
St1
n
6
5
1
0.056785 0.023 0252
6
= 0.010 720 8
0.010 720 8
h
0.023 025
0.010 720 8 0.023 025
=
= + 0.5 2 +
= 0 + 0.5
+
=
h
1/12
1/12
0.340 624 8
2 =
The answer is E.
Chapter 19
19.1
Lets forget about option pricing for now and focus on a random variable X.
Suppose we want to find out E (X), the mean of X, but we dont have an easy
formula to do so. For example, we have a random variable X = ez where z is a
standard normal random variable with mean 0 and variance 1.
R
R
1
E (X) = x (z) f (z) dz = exp (z) exp 0.5z 2 dz
2
R
1
We dont know how to do the integration exp (z) exp 0.5z 2 dz.
2
How can we calculate E (X)?
One approach is to create n instances of z ,
calculate
the corresponding
n
1 P
z
Xi . Based on the cenvalue of X = e , and find the sample mean =
n i=1
n
1 P
Xi is normally distributed with mean E () =
tral limit theorem, =
n i=1
n
n
P
P
1
1
1
Xi = (nE (X)) = E (X) and variance V ar () = 2 V ar
Xi =
E
n
n
n
i=1
i=1
1
V ar (X)
nV ar (X) =
. Since E () = E (X), we can calculate the mean of
2
n
n
the sample mean E () and use it to approximate the population mean E (X).
Lets come back to the example of estimating E (ez ). Im going to produce 10
sample means and the use the average of these 10 sample means as an estimate
to E (ez ). To produce each sample mean, Im going to randomly draw 10, 000
z 0 s (so I have 10, 000 values of ez to produce each sample mean). Another way
173
174
to describe this process is to say that Im going to have 10 trials with each trial
having 10, 000 samples.
Im going to do this in Microsoft Excel. The following table shows how to
calculate the sample mean for one trial.
1
2
3
4
5
6
7
8
......
10001
10002
10003
A
simulation i
1
2
3
4
5
6
7
...
10000
Total
sample mean
B
zi
0.042457
0.253383
0.017632
1.167590
0.273333
0.206665
0.305234
...
1.506862
C
Xi = ezi
0.958432
1.288376
1.017788
0.311116
1.314338
0.813292
1.356942
...
0.221604
16, 479.279610
1.647928
Lets walk through B2. Rand() is Excels formula for a random draw in a
uniform distribution over (0, 1). Norminv(p, , ) is Excels formula for the inverse normal distribution. For example, for a standard normal random variable
z N (0, 1), P (z 1.96) = N (1.96) = 0.975. Then Norminv(0.975, 0, 1) =
N 1
(0.975)
= 1.96. Another example. For a normal random variable Z
N 2, 0.32 , P (Z 2) = 0.5. Then Norminv(0.5, 2, 0.3) = 2.
NORMINV(RAND(),0,1) works this way. First, Rand() generates a number
from a uniform distribution over (0, 1). Say this random number is 0.4831. Then
Excel finds that P (z 0.042457) = 0.4831. Hence a =NORMINV(0.4831,0,1) =
0.042457.
Please note that Excel has a similar formula normsinv for calculating the
inverse normal random variable. The "s" in normsinv stands for standard. So
normsinv produces the inverse standard normal distribution, while norminv produces the inverse normal distribution with mean and standard deviation . In
other words, normsinv(p) =norminv(p, = 0, = 1). So normsinv(0.975) =norminv(0.975, 0, 1) =
1.96.
Similarly, the formula for B3 and C3 are:
B3=NORMINV(RAND(),0,1) = 0.253383
C3 = e0.253383 = 1. 288 376
B10001=NORMINV(RAND(),0,1) = 1.506862
C10001 = e1.506862 =
0.221 604
C10002=sum(C2:C10001)=16, 479.279610
19.1. EXAMPLE 1 ESTIMATE E E Z
175
The samplemean
is
n
P
1
16, 479.279610
=
Xi =
= 1.647928
10000 i=1
10000
i
1
2
3
...
10000
P
X
Trial 1
zi
0.042457
0.253383
0.017632
1.506862
Now I have 10
Trial
1
1.6479
2
1.6512
3
1.6163
4
1.6325
5
1.6821
6
1.6626
7
1.6656
8
1.5994
9
1.6459
10 1.6554
total 16.4589
Xi = ezi
0.958432
1.288376
1.017788
Trial 2
zi
0.985719
1.324269
0.157924
Xi = ezi
0.373171
0.265997
0.853915
0.221604
16, 479
1.673002
0.187683
16, 512
1.6479
...
Trial 10
zi
0.783098
0.186120
0.504305
1.177804
1.6512
Xi = ezi
0.456988
1.204567
1.655835
3.247236
16, 554
1.6554
E (X) = E () '
is P
P
P
2 !
2
i
1P
2i k ( i )2
k
1P 2
V ar () '
=
=
=
i
i
k1
k1
k1 k
k
10
2.709486 1.645892 = 0.000 591 2
9
p
176
1 0.95
=
2
Z is normal.
177
1
E Z = E (u1 + u2 + ... + u12 ) 6 = 12 6 = 0
2
1
V ar Z = V ar (u1 + u2 + ... + u12 ) = 12
=1
12
19.2
Example 2 Estimate
We decide to estimate using the Monte Carlo simulation (we can treat as
a random variable that happens to be a constant). Well estimate using a
classic "throwing the dart" method.
Imagine a square whose size is two units. The center of this square is the
origin (0, 0). Inside this square sits a unit circle x2 + y 2 = 1. The center of
this circle is also (0, 0). If we randomly throw a dart at the square, whats the
probability that the dart falls within the unit circle?
178
1.0
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
1.0
0.4
0.6
0.8
1.0
Of all the darts falling into the square (we ignore the darts falling out of the
square), the probability that the dart falls within the circle
is the area of the
12
. Then = 4P .
4
Next, we need to simulate darts falling in the first quadrant. Once again,
well do the simulation in Excel. We plan to produce 20 estimates of . For
each estimate, well use 10, 000 simulations. Heres how to produce one trial
(one trial=10, 000 simulations).
Create the first point (x1 , y1 ), where 0 < x1 < 1 and 0 < y1 < 1. We can
create x1 and y1 by randomly drawing two numbers from U (0, 1), a
uniform distribution over [0, 1].
If x21 + y12 1, then (x1 , y1 ) falls in the circle in the first quadrant.
Similarly, create the 2nd point (x2 , y2 ) by randomly drawing two numbers
from U (0, 1). Determine whether (x2 , y2 ) falls in the circle in the first
quadrant.
......
179
. Then = 4P .
n
of Excel.
B
C
xi
yi
0.2854
0.1694
0.5147
0.2787
0.1915
0.2699
0.6501
0.8420
0.0007
0.6580
0.1520
0.9772
0.3179
0.5713
...
...
0.5560 0.5679
D
x2i + yi2
0.1102
0.3426
0.1095
1.1317
0.4330
0.9780
0.4274
...
0.6317
E
Fall in the circle? (1=Yes, 0=No)
1
1
1
0
1
1
1
...
1
= 4 (0.7894) = 3. 157 6
Sample formulas in the above Excel spreadsheet are:
B2=rand()=0.2854
C2=rand()=0.1694
D2=B2^2+C2^2=0.1102
E2=if(D2>=1,1,0)=1
B3=rand()=0.5147
C3=rand()=0.2787
D3=B3^2+C3^2=0.3426
E3=if(D3>=1,1,0)=1
m=SUM(E2:E10001)=7, 894
We produce 19 more trials. The total number of trials is k = 20. Here is the
snapshot of Excel:
2
trial mi
pi
i
i
1
7,894
0.7894
3.1576
9.97043776
2
7,818
0.7818
3.1272
9.77937984
3
7,830
0.783
3.1320
9.80942400
4
7,752
0.7752
3.1008
9.61496064
5
7,879
0.7879
3.1516
9.93258256
6
7,879
0.7879
3.1516
9.93258256
7
7,778
0.7778
3.1112
9.67956544
180
8
9
10
11
12
13
14
15
16
17
18
19
20
total
0.7878
0.7891
0.7896
0.7844
0.788
0.792
0.7821
0.7825
0.787
0.7886
0.7822
0.7839
0.7762
3.1512
9.93006144
3.1564
9.96286096
3.1584
9.97549056
3.1376
9.84453376
3.1520
9.93510400
3.1680
10.03622400
3.1284
9.78688656
3.1300
9.79690000
3.1480
9.90990400
3.1544
9.95023936
3.1288
9.78938944
3.1356
9.83198736
3.1048
9.63978304
62.7856
197.10829728
P
62.7856
i
E =
=
= 3. 139 28
k
20
2 !
P
P
k
1
20 197.10829728
1
=
2i
i
3. 139 282 =
V ar '
k1 k
k
19
20
0.0003536 27
p
19.3
Monte Carlo simulation can be used to price European options, especially when
theres no simple formula for the option price such as the arithmetic Asian
option.
P
1 rT P
V STi , T = erT
e
V STi , T
n
t=1
t=1 n
n 1
i
P
V ST , T is the average terminal payo. So the price of the European
t=1 n
option is just the discounted value of the terminal payo.
V (S0 , 0) =
Were are going to produce 5 sample means. For each sample mean, well
use 10 simulations. So we have 5 trials with 20 simulations per trial. For each
trial, we need to generate the stocks terminal price at T = 0.25.
We generate hthe terminal stock price using
DM 19.3:
i
ST = S0 exp 0.5 2 T + T Z
Please note that the above formula produces the real world stock price at
T . If we use the real world terminal stock price to calculate the option price,
you have to use the real world discount rate, which is pathdependent (see DM
Table 19.1). Path dependent discount rates are dicult to calculate. To avoid
the diculty of finding the pathdependent discount rates, well want to live in
the risk neutral world where everything earns the riskfree rate. To move into
the risk neutral hworld, we just set = r andi change DM 19.3 into:
ST = S0 exp r 0.52 T + T Z
S0.25 = 41 exp 0.08 0.5 0.32 0.25 + 0.3 0.25Z = 41 exp (0.15Z + 0.008 75)
S0.25
Call payo
45.7901
5.7901
45.6185
5.6185
49.6034
9.6034
put payo
0.0000
0.0000
0.0000
182
4
1.50347
51.8234
11.8234
0.0000
5
0.62495
45.4251
5.4251
0.0000
6
1.89286
31.1369
0.0000
8.8631
7
0.57282
45.0712
5.0712
0.0000
8
0.84887
36.4154
0.0000
3.5846
9
0.34767
39.2586
0.0000
0.7414
10 0.34540
39.2720
0.0000
0.7280
Total
429.4146
43.3317
13.9171
Sample calculations.
If Z = 0.67830, S0.25 = 41 exp (0.15 0.67830 + 0.008 75) = 45. 790 1
The call payo is 45. 790 1 40 = 5. 790 1
The put payo is zero.
43.3317
= 4. 333 17
10
rT
So the call price is 4. 333 17e
= 4. 333 17e0.080.25 = 4. 247 4
13.9171
= 1. 391 71
10
0.080.25
= 1. 364 2
So the put price is 1. 391 71e
The average put payo at T is:
Now Im going to repeat this process 4 more times. This is my final result:
trials
call price
put price
1
4. 247 4
1. 364 2
2
3.9132
1.2391
3
0.5436
2.4875
4
1.6850
1.8342
5
2.4911
1.9250
The mean of the sample mean for the call price is:
4. 247 4 + 3.9132 + 0.5436 + 1.6850 + 2.4911
= 2. 58
5
Compare this with the correct price using the BlackScholes formula (see
DM page 377): C = 3.40
The mean of the sample mean for the put price is:
1. 364 2 + 1.2391 + 2.4875 + 1.8342 + 1.9250
= 1. 77
5
Compare this with the correct price using the BlackScholes formula (see
DM page 377): P = 1.61
The prices from the Monte Carlo simulations are o because the number of
simulations per trial is small.
After running 10 trials (5,000 simulations per trial), I got the following:
trials
call
call^2
put
put^2
1
3.4132
11.6502
1.5976
2.5523
3.3931
3.4085
3.4002
3.4019
3.4224
3.4089
3.4140
3.3998
3.4018
11.5134
11.6182
11.5614
11.5729
11.7129
11.6206
11.6551
11.5587
11.5725
1.6003
1.6007
1.6249
1.5815
1.6056
1.6158
1.6039
1.5879
1.6216
2.5611
2.5622
2.6402
2.5012
2.5778
2.6108
2.5725
2.5214
2.6297
184
19.4
Lets use the Monte Carlo method to calculate the price of an arithmetic call
option and an arithmetic put option. The inputs are (See DM page 629 Table
19.3):
S0 = 40
K = 40
= 0.3
r = 8%
=0
T = 0.25 (3 months)
The average stock price is the average stock prices at the end of Month 1,
Month2, and Month 3
Solution
The monthly interval is h = T /3 = 0.25/3
The stock prices (the real world price)
at the end of Month 1, 2, and 3 are:
end of Month 1:
Sh = S0 exp 0.5 2 h + hz1
end of Month 2:
S2h = Sh exp 0.5 2 h + hz2
end of Month 3:
S3h = ST = S2h exp 0.5 2 h + hz3
where z1 , z2 , and z3 are three separate random draws of the standard normal
distribution.
Since our goal is to calculate the option price, we dont need the real world
stock price. We just need the riskneutral stock price. We change the stocks
expected return into the risk free rate r. The risk neutral stock prices at the
end of Month 1,2, and 3 are:
Sh + S2h + S3h
3
C = erT max S K, 0
P = erT max K S, 0
185
The following snapshot of Excel shows one trial (5000 simulations per trial)
of the call/put price:
i
z1
Sh
z2
S2h
z3
S3h
S
1
0.8424 43.1529
0.258 44.2568 2.4041 36.0434 41.1510
2
0.066 40.3468 1.4415 35.7157 2.0925 29.8830 35.3152
3
0.5357 42.0218
0.1255 42.6051 0.8295 39.7677 41.4649
4
0.0014 40.1217
1.0027 43.8893 0.8446 40.9128 41.6413
5
1.598 34.9321 1.5906 30.5258
1.2832 34.2134 33.2238
6
1.2843 44.8364 0.6896 42.3605 0.3961 41.0516 42.7495
7 0.0712 39.8702
0.2363 40.8134
0.0147 40.9848 40.5561
8 1.0075 36.7649 0.5005 35.3082
0.6539 37.4745 36.5159
...
...
...
...
...
...
...
...
5000 0.0168 40.0585 1.2203 36.1464
0.6609 38.3874 38.1974
C pay
1.1510
0.0000
1.4649
1.6413
0.0000
2.7495
0.5561
0.0000
...
0.0000
P pay
0.0000
4.6848
0.0000
0.0000
6.7762
0.0000
0.0000
3.4841
...
1.8026
!
r
0.25
0.25
Sh = 40 exp 0.08 0 0.5 0.32
+ 0.3
0.8424 = 40 exp (0.075 87) =
3
3
43. 152 9
!
r
0.25
2 0.25
+ 0.3
0.258 =
S2h = 43. 152 9 exp 0.08 0 0.5 0.3
3
3
2
= 44. 256 8
43. 152 9 exp 2. 526 01
10
!
r
0.25
2 0.25
S3h = 44. 256 8 exp 0.08 0 0.5 0.3
+ 0.3
(2.4041) =
3
3
44. 256 8 exp (0.205 28) = 36. 0434
43. 152 9 + 44. 256 8 + 36. 0434
S=
= 41. 151 0
3
index
1
2
3
4
5
6
7
8
...
5000
sum
Call payo
1.1510
0.0000
1.4649
1.6413
0.0000
2.7495
0.5561
0.0000
...
0.0000
10, 111.9410
put payo
0.0000
4.6848
0.0000
0.0000
6.7762
0.0000
0.0000
3.4841
...
1.8026
7, 390.6006
C
1.1282
0.0000
1.4359
1.6088
0.0000
2.6951
0.5451
0.0000
...
0.0000
9, 911.7119
P
0.0000
4.5920
0.0000
0.0000
6.6420
0.0000
0.0000
3.4151
...
1.7669
7, 244.2548
C2
1.2728
0.0000
2.0618
2.5882
0.0000
7.2636
0.2971
0.0000
...
0.0000
61, 373.4417
P2
0.0000
21.0865
0.0000
0.0000
44.1162
0.0000
0.0000
11.6629
...
3.1219
32, 916.8904
186
index
Call payo put payo
C
sum 10, 111.9410 7, 390.6006 9, 911.7119
The average call price of these 5000 simulations:
erT
or
P
7, 244.2548
C2
61, 373.4417
P2
32, 916.8904
1 P
10111.9410
call payo= e0.08(0.25)
= 1. 982 34
5000
5000
9911.7119
1 P
C=
= 1. 982 34
5000
5000
erT
or
1 P
7390.6006
put payo= e0.08(0.25)
= 1. 448 9
5000
5000
1 P
7244.2548
P =
= 1. 448 9
5000
5000
P 2 !
C
5000
1 P 2
1
n
=
C
61373.4417 1. 982 342 =
n1 n
n
5000 1 5000
8. 346 7
P 2 !
2 !
n
5000
P
1 P 2
1
7244.2548
=
P
32916.8904
n1 n
n
5000 1 5000
5000
= 4. 485 1
The following is the snapshot of Excel for 30 trials (5000 simulations per
trial) for the arithmetic European call and put prices:
C
1.9823
1.9409
1.9467
2.0098
1.9894
1.9515
1.9825
2.0462
2.1270
1.9583
1.9878
1.9664
2.0147
1.9664
1.9915
1.9504
1.9414
1.9804
1.9666
2.0276
1.9940
1.9307
1.9221
1.9710
1.9185
1.9529
1.9791
1.9776
1.9713
1.9689
59.3139
P
1.4489
1.4486
1.4920
1.4211
1.4447
1.4886
1.4813
1.4408
1.3565
1.4675
1.4691
1.4496
1.3785
1.4327
1.4253
1.4592
1.4800
1.4186
1.4724
1.3779
1.4217
1.4815
1.4903
1.4491
1.4526
1.4638
1.4538
1.4676
1.5092
1.4640
43.5069
C2
3.929513
3.767093
3.789641
4.039296
3.957712
3.808352
3.930306
4.186934
4.524129
3.834939
3.951349
3.866729
4.059016
3.866729
3.966072
3.804060
3.769034
3.921984
3.867516
4.111162
3.976036
3.727602
3.694468
3.884841
3.680642
3.813818
3.916837
3.910902
3.886024
3.876567
117.3193
187
P2
2.099311
2.098442
2.226064
2.019525
2.087158
2.215930
2.194250
2.075905
1.840092
2.153556
2.158255
2.101340
1.900262
2.052629
2.031480
2.129265
2.190400
2.012426
2.167962
1.898608
2.021231
2.194842
2.220994
2.099891
2.110047
2.142710
2.113534
2.153850
2.277685
2.143296
63.1309
For the remaining part of the calculation, all you need to know is the following:
Trial
C
P
C2
P2
sum 59.3139 43.5069 117.3193 63.1309
188
Ci
P 2
n
1
1 n=30
i=1
2
= 30
Ci
V ar C =
=
117.3193
1.
977
1
n 30 1 30
n 1 n
i=1
n=30
P
0.001 778
n=30
2
P
P
i
P 2
1
30
n
1 n=30
i=1
2
Pi
63.1309 1. 450 23 =
V ar P =
=
n 1 n
n 30 1 30
i=1
0.001 237 5
5,000
P
1
1
1
V ar P =
V ar
Pi =
5, 000V ar (P ) =
V ar (P ) =
2
2
5, 000
5, 000
5000
i=1
4. 4668
= 0.000 89
5000
189
S0 = 40
K = 40
= 0.3
r = 8%
=0
T = 0.25 (3 months)
The average stock price is the average stock prices at the end of Month 1,
Month2, and Month 3
z2
1.2203
S2h
36.1464
z3
2.4041
2.0925
0.8295
0.8446
1.2832
0.3961
0.0147
0.6539
...
0.6609
z3
0.6609
S3h
36.0434
29.8830
39.7677
40.9128
34.2134
41.0516
40.9848
37.4745
...
38.3874
S3h
38.3874
S
40.9831
35.0508
41.4466
41.6101
33.1662
42.7209
40.5532
36.5047
...
38. 163 6
S
38. 163 6
C pay
0.9831
0
1.4466
1.6101
0
2.7209
0.5532
0
....
0
9, 911.3146
C pay
0
P pay
0
4.9492
0
0
6.8338
0
0
3.4953
...
1. 836 4
7, 528.5134
P pay
1. 836 4
190
put payo
0
4.9492
0
0
6.8338
0
0
3.4953
...
1. 836 4
7, 528.5134
C
0.9636
0
1.418
1.5782
0
2.667
0.5422
0
...
0
9, 715.0581
P
0
4.8512
0
0
6.6985
0
0
3.4261
...
1. 800 0
7, 379.4385
C2
0.9285
0
2.0107
2.4907
0
7.1129
0.2940
0
P2
0
23.5341
0
0
44.8699
0
0
11.7382
0
59, 337.3153
3.24
33, 852.0593
Sample calculation.
The 1st simulation.
Call price: 0.9831e0.080.25 = 0.963 6
Put price: 0 e0.080.25 = 0
The 5000th simulation.
Call price: 0
Put price: 1. 836 4e0.080.25 = 1. 800 0
The average call price of these 5000 simulations:
1 P
10111.9410
erT
call payo= e0.08(0.25)
= 1. 982 34
5000
5000
or
1 P
9911.7119
C=
= 1. 982 34
5000
5000
C2
59, 337.3153
P2
33, 852.0593
1 P
9715.0581
C=
= 1. 943 0
5000
5000
The estimated
variance of the call
price per simulation is:
P 2 !
C
5000
1 P 2
1
n
=
C
59337.3153 1. 943 02 =
n1 n
n
5000 1 5000
8. 093 8
1 P
7528.5134
put payo= e0.08(0.25)
= 1. 475 9
5000
5000
1 P
7379.4385
P =
= 1. 475 9
5000
5000
The estimated
variance of the put price per
simulation is:
1
5000
33852.0593 1. 475 92 = 4. 593 0
5000 1 5000
Next is the snapshot of the 30 trials (1 trial=5,000 simulations):
i
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
sum
C
1. 943 0
1.9001
1.9080
1.9698
1.9506
1.9112
1.9436
2.0058
2.0854
1.9191
1.9487
1.9281
1.9748
1.9272
1.9519
1.9116
1.9039
1.9424
1.9274
1.9867
1.9535
1.8924
1.8845
1.9328
1.8804
1.9133
1.9399
1.9398
1.9318
1.9279
58.1356
P
1.4759
1.4754
1.5200
1.4476
1.4727
1.5171
1.5091
1.4675
1.3830
1.4958
1.4959
1.4775
1.4058
1.4608
1.4523
1.4863
1.5072
1.4453
1.5007
1.4049
1.4483
1.5100
1.5182
1.4771
1.4809
1.4916
1.4814
1.4947
1.5376
1.4921
44.3327
C2
3.775249
3.610380
3.640464
3.880112
3.804840
3.652685
3.777581
4.023234
4.348893
3.682945
3.797432
3.717570
3.899835
3.714100
3.809914
3.654215
3.624835
3.772918
3.714871
3.946977
3.816162
3.581178
3.551340
3.735716
3.535904
3.660717
3.763212
3.762824
3.731851
3.716798
112.7048
P2
2.178281
2.176805
2.310400
2.095546
2.168845
2.301592
2.277383
2.153556
1.912689
2.237418
2.237717
2.183006
1.976274
2.133937
2.109175
2.209088
2.271652
2.088892
2.252100
1.973744
2.097573
2.280100
2.304931
2.181824
2.193065
2.224871
2.194546
2.234128
2.364214
2.226362
65.5497
191
192
C2
P2
112.7048 65.5497
58.1356
The estimated call price is:
= 1. 937 85
30
44.3327
= 1. 477 76
The estimated put price is:
30
trial
sum
C
58.1356
P
44.3327
The estimated
variance of the call price
per trial:
30
1
2
112.7048 1. 937 85 = 0.001618
30 1 30
The estimated
variance of the putprice per trial:
30
1
65.5497 1. 477 762 = 0.001 257 3
30 1 30
The standard deviation is:
The estimated variance of the call per trial can also be calculated as follows:
For each trial, the average put price of the 5, 000 simulations is used to
P
1 5,000
Pi .
estimate the put price. P =
5, 000 i=1
So the variance of the
priceper trial is:
put
5,000
P
1
1
4. 593 0
V ar P =
Pi =
V ar
V ar (P ) =
= 0.000 918 6
2
5, 000
5000
5000
i=1
The standard deviation is:
V ar
n
P
Ci
i=1
1
V ar (C)
nV ar (C) =
n2
n
Set
5
V ar (C)
= 0.02. We have:
n
193
4.5930
4. 593 0
= 11482.
= 0.02 or n =
n
0.022
19.5
19.5.1
V ar (V )
nV ar (V )
=
2
n
n
V
The standard deviation of the sample mean is V = , where n is the numn
ber of the simulations and V is the standard deviation of the option price per
simulation. To decrease V , we need to increase n by doing more simulations.
Doing more simulations costs time.
However, there are techniques out there to reduce V without increasing n.
One method is called the control variate method. It goes like this.
Suppose we have two random similar variables X and Y . We need to calculate E (X) and E (Y ). We can calculate E (X) easily because theres a formula
for E (X). E (Y ), on the other hand, doesnt have a formula and needs to be
calculated through the Monte Carlo simulation. Since X and Y are similar, we
expect that our errors in estimating E (X) are similar to our errors in estimating
E (Y ):
V1 + V2 + ... + Vn
n
E (X) X E (Y ) Y
In the above formula, E (X) and E (Y ) are the true means of X and Y ;
X and Y are the estimated means of X and Y based on the Monte Carlo
simulation.
Since our goal his to find E (Y
i ), we
arrangethe above formula into:
E (Y ) Y + E (X) X = Y X + E (X)
estimate as Y = Y X + E (X).
194
the old estimate E (Y ) Y ? It turns out the new estimate often has lower
The
variance
estimate
is: i
of the
hnew
V ar Y = V ar Y X + E (X) = V ar Y X = V ar X +
V ar Y 2Cov X ,Y
The above formula holds because E (X) is a constant.
Now lets go through an example. Theres no formula for the arithmetic
European call option price; theres a formula for the geometric European call
option price. So well use the Monte Carlo simulation to estimate the arithmetic
European call option. Well use the geometric European price as the control
variate (i.e. a dummy variable).
Let AC=arithmetic European call price
price.
AC = AC GC + E (GC)
4 0.32 4 7
1
2
=
0.08 + 0 + 0.5 0.32
2
= 0.03 333
2
3
3
3
6
Using DM 14.19, we have:
0.3
3
195
47
= 0.216 025
6
1
40
1
S
ln
+ r + ( )2 T
+ 0.08 0.03 333 + 0.216 0252 0.25
ln
K
2
40
2
d1 =
=
=
0.216 025 0.25
T
0.162 026
Now we have:
AC = AC GC + 1. 938 5
196
(AC GC)2
0.001544
0.001665
0.001498
0.001600
0.001505
0.001624
0.001513
0.001632
0.001731
0.001537
0.001529
0.001467
0.001592
0.001537
0.001568
0.001505
0.001406
0.001444
0.001537
0.001673
0.001640
0.001467
0.001414
0.001459
0.001452
0.001568
0.001537
0.001429
0.001560
0.001681
0.046313
AC
GC AC GC
59.3139 58.1356
1.1783
59.3139
=
= 1. 977 13
30
(AC GC)2
0.046313
i
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
sum
AC
1.9823
1.9409
1.9467
2.0098
1.9894
1.9515
1.9825
2.0462
2.1270
1.9583
1.9878
1.9664
2.0147
1.9664
1.9915
1.9504
1.9414
1.9804
1.9666
2.0276
1.9940
1.9307
1.9221
1.9710
1.9185
1.9529
1.9791
1.9776
1.9713
1.9689
59.3139
GC
1.9430
1.9001
1.9080
1.9698
1.9506
1.9112
1.9436
2.0058
2.0854
1.9191
1.9487
1.9281
1.9748
1.9272
1.9519
1.9116
1.9039
1.9424
1.9274
1.9867
1.9535
1.8924
1.8845
1.9328
1.8804
1.9133
1.9399
1.9398
1.9318
1.9279
58.1356
So
i
sum
AC
58.1356
GC =
= 1. 937 85
30
Hence the updated estimated price of the geometric European call option is:
197
1.1783
= 0.03927 67
30
= 0.03927 67 + 1. 938 5 = 1. 977 78
AC GC =
AC
V ar AC = V ar AC GC + 1. 938 5 = V ar AC GC =
1. 145 7 106
30
30 1
V ar GC = 0.001618
We can see that V ar AC is much smaller than V ar AC .
lower variance than the variance of the original estimate Y . Boyle recommends
the following newestimate:
Y = Y + E (X) X
If we use the textbook
notation, the above new estimate is DM 19.10:
A = A + G G
V ar (A ) = V ar A + G G = V ar A G = V ar A + 2 V ar G
2Cov A, G
Cov A, G
.
The textbook says that V ar (A ) is minimized if we set =
V ar G
How do we find ? Typically, we perform a small number of Monte Carlo
simulations, run regression of DM 19.10, and estimate . Then we apply the
estimated to DM 19.10, run more simulations, and calculate A .
This is all you need to know about the Boyles improved control variate
method.
19.6
1
0.046313 0.03927 672
30
198
standard normal distributions. Since z1 , z2 ,...,zn are random draws from the
standard normal distribution, z1 , z2 ,...,zn are also random draws from the
standard normal distribution. Next, we calculate two samples means, X 1 (using
z1 , z2 ,...,zn ) and X 2 (using z1 , z2 ,...,zn ). Then we can estimated E (X) as
the average of X 1 and X 2 :
X1 + X2
E (X)
2
This method is called the antithetic variate method.
Heres an example. We want to estimate E (ez ) where z is the standard
normal random variable. We have generated the following 10 standard normal
random variables:
i
zi
1 0.0183
2
2.0478
3 0.4849
4 0.6583
5 0.4666
6 0.3757
7
1.1392
8
2.1566
9
0.1096
10 1.5389
This is
i
1
2
3
4
5
6
7
8
9
10
sum
X=
24.276548
= 2. 427 654 8
10
Y =
13.745027
= 1. 374 502 7
10
E (ez )
19.7
199
Stratified sampling
19.7.1
Importance sampling
In the importance sampling, we perform simulations from a conditional distribution, not from the original distribution. For example, we want to estimate
the price of an option that is deep out of the money. If we perform simulations
from the original distribution, then most of the simulated payos will be zero.
This is a waste of our time. To make the simulations more ecient, well draw
random numbers from the conditional distribution where the payo is not zero.
This is all you need to know about the importance sampling.
The textbook also mentioned Latin hypercube sampling and low discrepancy
sequences. Since the textbook merely mentioned these terms without providing
much explanation, I dont think SOA expects you to know much about term.
Skip these terms and move on.
19.8
Sample problems
Problem 1
200
You are simulating the standard normal random variable z by taking random
draws from a uniform distribution over (0, 1). Let a represent the simulated
value of z. Calculate P (z a). You are given:
i
1
2
3
4
5
6
7
8
9
10
11
12
ui
0.3763
0.1349
0.414
0.0405
0.5225
0.0423
0.2041
0.9282
0.6792
0.3368
0.1535
0.157
Solution
P
ui = 0.3763+ 0.1349 + 0.414 + 0.0405+ 0.5225 + 0.0423 + 0.2041+ 0.9282 +
0.6792 + 0.3368 + 0.1535 + 0.157 = 3. 989 3
P
The simulated value is a = ui 6 = 3.9893 6 = 2. 010 7
P (z a) = N (2. 010 7) = 1 N (2. 010 7) = 0.022
Problem 2
You are simulating E (ez ) by taking random 10 draws from a uniform distribution over (0, 1). Calculate the simulated value of E (ez ). You are given:
i
1
2
3
4
5
6
7
8
9
10
ui
0.878
0.762
0.069
0.8
0.048
0.22
0.178
0.661
0.191
0.258
Solution
201
zi = N 1 (ui )
1.17
0.71
1.48
0.84
1.66
0.77
0.92
0.42
0.87
0.65
xi = ezi
3.221993
2.033991
0.227638
2.316367
0.190139
0.463013
0.398519
1.521962
0.418952
0.522046
11.314619
11.314619
The estimated value is E (ez ) =
= 1. 131 461 9
10
Sample calculation. u1 = 0.878. Look at the normal table. You see that
roughly P (z < 1.17) = 0.878 so z1 = 1.17.
i
1
2
3
4
5
6
7
8
9
10
sum
ui
0.878
0.762
0.069
0.8
0.048
0.22
0.178
0.661
0.191
0.258
202
ui U (0, 1)
0.6515
0.8839
0.7621
0.3922
0.1748
2
ST = S0 exp 0.5 2 T + T z = 100e(0.060.020.50.3 )0.5+0.3 0.5z =
0.5z
100e0.002 5+0.3
i
1
2
3
4
5
Total
u1 U (0, 1)
0.6515
0.8839
0.7621
0.3922
0.1748
zi = N 1 (u1 )
0.39
1.19
0.71
0.27
0.94
528.6276
= 105. 73
5
Please note that the risk free rate r is not needed for solving this problem.
Problem 4
You are simulating the price of an arithmetic average stock price European
call option and put option. You are given:
Stock prices are lognormally distributed
S0 = 40
K = 40
= 0.06
r = 0.08
=0
T =1
= 0.3
203
+ S2/3
+ S1
S1/3
rT
e
max
K, 0
3
stock price at t.
St is the riskneutral
2
St = Sth exp r 0.5 2 h + hz = Sth e(0.0800.50.3 )h+0.3 hz
u1
0.4828
z1
0.04
S1/3
40.1900
u2
0.6177
z2
0.3
S2/3
42.8303
u3
0.9345
z3
1.51
S1
56.2864
Sample calculation.
2
S = 42.8303e(0.0800.50.3 )1/3+0.3 1/3(1.51) = 56. 286 4
1
The simulated value of the call price is: e0.081 6. 435 6 = 5. 940 8
The simulated value of the put price is: e0.081 0 = 0
Problem 5 (spring 2007 Exam C #19)
The price of a non dividendpaying stock is to be estimated using simulation.
It is known that:
The price St follows the lognormal distribution
S0 = 50, = 0.15, and = 0.30.
204
Using the following uniform (0, 1) random numbers and the inversion method,
three prices for two years from the current date are simulated 0.9830, 0.0384, 0.7794.
Calculate the mean of the three simulated prices.
(A) Less than 75
(B) At least 75, but less than 85
(C) At least 85,
but less than 95
(D) At least 95, but less than 115
(E) At least 115
Solution
ST = S0 exp
2z
50e0.21+0.3
2
0.5 2 T + T z = 50e(0.1500.50.3 )2+0.3 2z =
zi = N 1 (u1 )
2. 12
1. 77
0.77
ST = 50e0.21+0.3 2z
22.12
50e0.21+0.3
= 151. 63
u1 U (0, 1)
0.9830
0.0384
0.7794
Chapter 20
Introduction
206
http://en.wikipedia.org/wiki/Brownian_motion
To help you experiment with the Brownian motion, I designed a spreadsheet
titled "Simulate Brownian Motion." Download this spreadsheet.
My spreadsheet uses the following Excel functions:
Rand() returns a random number equal to or greater than 0 but less than 1.
In other words, Rand() simulates a random variable uniformly distributed
over [0, 1).
NORMINV(probability,mean,standard_dev) returns the inverse of the normal cumulative distribution for the specified mean and standard deviation.
20.1.1
Big picture
When studying the Brownian motion and Itos lemma, remember the following
big picture. The BlackScholes option pricing formula Equation 12.1 relies on
the BlackScholes PDE (Equation 12.24). Equation 12.24 assumes the following
price model for a riskfree asset and a risky asset (i.e. stock):
The price of a risk free asset (i.e. the savings account or a bond) is
B (t) = ert
(20.1)
S (t) = S (0) e
1 2
t+ tY (t)
(20.2)
(Textbook 20.1)
207
20.2
Brownian motion
20.2.1
Stochastic process
20.2.2
Consider a particle that jumps, at discrete times, up or down along the vertical
line. At t = 0 the particle is at position zero. After each hlong time period,
the particle jumps up or down by a constant distance of k and with equal
probability of 0.5. That is, at t = h, 2h, 3h, ..., nh, the particle either moves
up by k or moves down by k, with up and down movements having an equal
probability of 0.5. Let Z (t) represent the height of the article from the position
zero at time t. Clearly Z (0) = 0. We like to find Z (T ), the height of the article
at time T = nh.
The particles height
4k
3k
2k
k
0
0
k
time 0
2k
k
2k
2h
0
k
3k
3h
k
4k
4h
Lets walk through the above table. At t = 0 the particle is at the position
zero. At t = h, the particles height is either k or k. At t = 2h, the k node
either goes up to 2k or goes down to 0. Similarly, the k node either goes up
to 0 or goes down to 2k. So on and so forth.
The jump at t = h is: Z (h) Z (0) = Y (h) k
Here Y (h) is a direction indicator. If Y (h) = 1, then the particle moves up
by k; if Y (h)
= 1, then the particle moves down by k:
1 Probability 0.5
Y (h) =
1 Probability 0.5
208
k2
=1
k= h
h
Now we have:
(20.3)
k2
T = T:
h
Please note that another way to specify the model is treat Y (ih) as a random
draw of a standard normal random variable (instead of a binomial random
variable):
209
(20.4)
My spreadsheet for simulating the Brownian motion uses both Equation 20.3
and Equation 20.4.
The random process Z (t) as n is called the Brownian motion or the
Wiener process.
Next, lets formally define the Brownian motion.
Definition 20.2.1.
A stochastic process Z (t) is a Brownian motion or a Wiener process if
1. Z (0) = 0 . Brownian motion starts at zero (this is merely for our convenience).
2. Z (t + h) Z (t) is normally distributed with mean 0 and variance h. This
means that the increments over a time interval h is normally distributed
with mean 0 and variance h. This stands true no matter how small or big
h is.
3. Z (t + s1 ) Z (t) is independent of Z (t) Z (t s2 ) where s1 , s2 > 0.
4. Z (t) is continuous.
20.2.3
Martingale
210
20.2.4
4. Cov [Z (s) , Z (t)] = min (s, t). The covariance of two Brownian motions is
the shorter time interval.
0
if n is odd
5. The higher moments of Z (t) is: E [Z n (t)] =
tn/2 (n 1) (n 3) ...1 if n is even
Lets look at the first property.
If you look at the simulation of Brownian motion over the internet, youll
find that the Brownian motion is always continuous yet its not dierentiable
anywhere. Can you imagine that a function is continuously anywhere yet dierentiable nowhere? If I hadnt studied the Brownian motion, I would have never
thought that such a function exists.
We can explain the nondierentiality using the following equation (its textbook Equation 20.4):
dZ (t) = Y (t) dt
(20.5)
211
EX = xf (x) dx > 0 (since x 0, its mean must be positive)
Then lim (Z (h) Z (0)  + Z (2h) Z (h)  + ... + Z (nh) Z [(n 1) h] )
n
0.000001
tiny interval t, t is much large than t. For example,
= 1, 000.
0.000001
Hence during any short interval, the Brownian motion can move up or down by
an infinitely large amount.
In contrast, for a continuously dierentiable function y = f (t), we have:
y = f 0 (t) t 0 as t 0
Lets get an intuitive
feel
of the 3rd property. On average, {Z [(i + 1) h]
2
2
approaches
E
X
.
Z (ih)}
E X 2 = E 2 (X) + V ar (X) = 02 + h = h
2
2
2
lim [Z (h) Z (0)] + [Z (2h) Z (h)] + ... + (Z (nh) Z [(n 1) h]) apn
proaches
h + h + ... + h = nh = T
In contrast, the 2nd order variation of a dierentiable function is zero. For
example, we can find the 2nd order variation of the function y = x is zero.
Divide the interval [0, T ] into [0, h],[h, 2h],...,[(n 1) h, nh = T ]
2
T
2
2
2
2
lim {(h 0) + (2h h) +... +[nh (n 1) h]} = nh = lim n
=
n
n
n
0
It can be proven that the quadratic variation of any continuously dierentiable function is zero. The quadratic variation of any continuously dierentiable
function is zero because such a function is roughly linear at any point. For a
continuously dierentiable function y = f (t), we have:
2
y = f 0 (t) t
(y)2 = [f 0 (t)] (t)2
2
2
For a tiny interval t, (t) 0 must faster than t 0. Hence (y) 0
212
2
In contrast, for a Brownian motion we have Z (t) = Y (t) t and [Z (t)] =
P
P
2
2
[Y (t)] t t. Hence
[Z (t)] t = T
Brownian motion Property 3 reconfirms the idea that Brownian motion is
not dierentiable anywhere. If its dierentiable, then its second order variation
would be zero.
Tip 20.2.1. Just memorize Equation 20.5 dZ (t) = Y (t) dt. This equation
tells you that the Brownian motion Z (t) is not dierentiable anywhere, its second order variation is t, and its first order variation is infinite.
Tip 20.2.2. To get an intuitive feel of the equation dZ (t) = Y (t) dt, imagine
you are looking at the Brownian motion under a magnifying class. If you zoom
in on the Brownian motion by shrinking the time interval dt, no matter how
much you reduce dt, youll see a jigsaw. In comparison, if you zoom in on a
continuously dierentiable function such as y = t2 , youll see a straight line.
Please note that Derivatives Markets explains Property 2 and 3 using the
following formula:
2
2
2
lim {[Z (h) Z (0)] + [Z (2h) Z (h)] + ... + (Z (nh) Z [(n 1) h]) } =
n
nh = T
The problem with this explanation is that it works if we treat Y [(i + 1) h]
as binomial random variable whose value is 1. Such explanation wont work if
we treat Y [(i + 1) h] as a random draw of a standard normal random variable.
The explanation I provided here works no matter if you treat Y [(i + 1) h] as a
binomial random variable or a standard normal random variable.
Lets look at Property 4. Suppose s t
Cov [Z (s) , Z (t)] = Cov{Z (s) , Z (s) + [Z (t) Z (s)]}
Using the formula Cov (a, b + c) = Cov (a, b) + Cov (a, c), we get:
Cov{Z (s) , Z (s)+[Z (t) Z (s)]} = Cov{Z (s) , Z (s)}+Cov{Z (s) , [Z (t) Z (s)]}
Cov{Z (s) , Z (s)} = V ar [Z (s)] = s
Since Z (s) and [Z (t) Z (s)] are independent (Brownian motion definition
Point #3), Cov{Z (s) , [Z (t) Z (s)]} = 0
Cov [Z (s) , Z (t)] = s = min (s, t)
Property 5 is based on the moment formula for a standard normal random
variable
0
if n is odd
E (n ) =
(20.6)
(n 1) (n 3) ...1 if n is even
We can find the nth moment of a random variable X using the moment
generating function (MGF):
213
dn
E (X ) =
MX (t)
dtn
n
(20.7)
t=0
The MGF of a normal random variable X with mean and the standard
deviation is:
1
tX
t+ 2 t2
2
MX (t) = E e
=e
(20.8)
1 2
t
M (t) = e 2
Using Equation 20.7, you can verify that Equation 20.6 holds.
Since Z (t) is a normal random variable with mean 0 and variance t, then
Z (t)
is a standard normal random variable. Hence
t
0
if n is odd
(20.9)
E [Z n (t)] =
n/2
(n 1) (n 3) ...1 if n is even
t
Example 20.2.1. Calculate P [Z (3) > 1]
Z (3) is a normal random variable with mean
0 andvariance 3.
10
P [Z (3) > 1] = 1 P [Z (3) 1] = 1
= 1 (0.577 35) =
3
0.281 9
Example 20.2.2. Calculate P [Z (1) 0 Z (2) 0]
Z (1) is a normal random variable with mean 0 and variance 1. Let X = Z (1)
Z (2) = Z (1) + [Z (2) Z (1)]
Z (2) Z (1) is a normal random variable with mean 0 and variance 1.
Z (2) = Z (1) + [Z (2) Z (1)] 0
[Z (2) Z (1)] Z (1)
Let Y = Z (2) Z (1).
X and Y are independent.
P [Z (1) 0 Z (2) 0] = P (X 0 Y X)
To have X 0 Y X, we first fix X at a tiny interval (x, x + dx)
where < x < 0. Next, we set Y < x. Then we are guaranteed to have
X 0Y X. Let f (x) and (x) represent the probability density function
(pdf) and the cumulative density function (cdf) of a standard normal random
variable.
R0
R0
P (X 0 Y X) = P (x < X < x + dx) P (Y < x) = [f (x) dx] P (Y < x)
However, f (x) dx = d (x) and P (Y < x) = (x) = 1 (x)
R0
R0
R0
[f (x) dx] P (x) = [d (x)] [1 (x)] = [1 (x)] d (x) =
R0
[1 (x)] d (x)
0
1 2
= (x) (x)
2
214
1 2
= [ (0) ()]
(0) 2 ()
2
#
" 2
1
1 1
2
=
0
0
2
2 2
2
1 1 1
3
=
=
2 2 2
8
Example 20.2.3. Calculate Cov [Z (5) , Z (2)]
Cov [Z (5) , Z (2)] = min (5, 2) = 2
20.2.5
(20.10)
dX (t) = dt + dZ (t)
(20.11)
Equation 20.10 and 20.11 are called arithmetic Brownian motion. is the
instantaneous mean per unit of time; is the instantaneous standard deviation
per unit of time.
Equation 20.10 and 20.11 indicate that X (T )X (0) is normally distributed.
Its mean and variance are:
E [X (T ) X (0)] = E [T + Z (T )] = T + E [Z (T )] = T + 0 = T
V ar [X (T ) X (0)] = V ar [T + Z (T )] = V ar [Z (T )] = 2 V ar [Z (T )] =
2T
The textbook lists the major properties and weaknesses of Equation 20.11.
Major properties:
1. X (t) is normally distributed.
215
dX (t)
be rewritten as
=
dt +
dZ (t). This indicates that
,
X (t)
X (t)
X (t)
X (t)
the percentage return on the stock depends on the stock price X (t). However,
in reality, we think that the stock return on average shouldnt depend on the
stock price. In other words, instead of Equation 20.11, we like to see
dX (t)
= dt + dZ (t)
X (t)
(20.12)
(20.13)
or
Equation 20.12 and 20.13 are called the geometric Brownian motion.
20.2.6
OrnsteinUhlenbeck process
We can modify Equation 20.11 to allow for mean reversion. Its reasonable for
us to assume that the stock price or the interest rate will revert to the mean.
For example, if the stock price is too high, then it might go down; if the stock
price is too low, it might go up. We modify the drift term in Equation 20.11:
dX (t) = [ X (t)] dt + dZ (T )
(20.14)
216
(20.15)
20.3
(i1)h
R1 2
x dx
0
2
2
1 (n 1) (n) (2n 1)
x2 dx = lim 3
n
n
6
1
1
1
1
= lim
1
(1) 2
=
n 6
n
n
6
R ih
There are other ways to approximate (i1)h x2 dx. For example, the area of
function x2 over the interval [(i 1) h, ih] is roughly the area of the rectangular
with height (ih)2 and width ih (i 1) h = h.
R ih
2
x2 dx (ih) h = i2 h3
(i1)h
0
x2 dx = lim
R ih
i=1 (i1)h
Pn
x2 dx = lim
Pn
i=1
217
n (n + 1) (2n + 1) 3
h
n
6
i2 h3 = lim
1
n (n + 1) (2n + 1) 1
= lim
=
n
6
n3 R 6
ih
The 3rd way to approximate (i1)h x2 dx is to take the average height of
the rectangular. So the area of function x2 over the interval [(i 1) h, ih] is
2
2
(i 1) h2 + (ih)
and width
roughly the area of the rectangular with height
2
ih (i 1) h = h.
2
2
2
(i 1) h2 + (ih)
(i 1) + i2 2
=
h
2
2 #
"
2
R ih
(i 1) + i2 2
2
x
dx
h h
(i1)h
2
"
#
2
2
R1 2
Pn R ih
P
(i
1)
+
i
n
2
x dx = lim
lim
h2 h
i=1 (i1)h x dx = n
i=1
0
n
2
i
Pn h
Pn 2 2
1
1
2 2
h
(i
1)
h + lim
i h h
=
lim
i=1
n
2
2 n i=1
1 1
1 1
1
=
+
=
2 6
2 6
6
It seems natural that we extend this logic of deterministic integration to a
define a stochastic integration.
Suppose we partition [a, b] into a = t0 < t1 < t2 < ... < tn = b. We can
make the partition intervals [t0, t1 ], [t1, t2 ], ..., [tn1, tn ] have the same length
ba
tk+1 tk =
. We can also have the partition intervals [t0, t1 ], [t1, t2 ], ...,
n
[tn1, tn ] have dierent lengths.
Rb
Pn1
g (t) dZ (t) = lim
k=0 g (tk ) [Z (tk+1 ) Z (tk )]
a
n
Definition 20.3.1.
Suppose g (t) is a simple process, meaning that g (t) is piecewiseconstant
Rb
but may have jumps at a = t0 < t1 < t2 < ... < tn = b. If a E g 2 (t) dt < ,
Rb
then the stochastic integral a g (t) dZ (t) is defined as
Z
g (t) dZ (t) =
n1
X
k=0
(20.16)
Rb
In the above definition, a E g 2 (t) dt < is the sucient condition for
Pn1
k=0 g (tk ) [Z (tk+1 ) Z (tk )] to exist.
Example 20.3.1. Calculate
Solution.
RT
0
dZ (t)
218
Here g (t) = 1.
RT
R T 2
E 1 dt = T < . So 0 dZ (t) exists.
0
RT
Pn1
dZ (t) = k=0 [Z (tk+1 ) Z (tk )] = Z (T )
0
Example 20.3.2.
1 if 0 t 1
2 if 1 < t 2
X (t) =
3 if 2 < t 3
R3
Calculate 0 X (t) dZ (t)
Solution.
Dividend [0, 3] into (0, 1), (1, 2), and (2, 3). Then X (t) is constant during
the Rinterval and jumps at t = 1 and t = 2. Hence
3
X (t) dZ (t)
0
= X (0) [Z (1) Z (0)] + X (1) [Z (2) Z (1)] + X (2) [Z (3) Z (2)]
= 1 [Z (1) Z (0)] + 2 [Z (2) Z (1)] + 3 [Z (3) Z (2)]
= 1Z (1) + 2 [Z (2) Z (1)] + 3 [Z (3) Z (2)]
= 3Z (3) Z (2) Z (1)
If g (tk ) is not a simple process, then we define the stochastic integral as
follows.
If E lim (X a)2 = 0, we say that X approach a in mean square.
n
Explain why the sample mean approaches the population mean in mean
square.
Suppose we take n random samples X1 , X2 ,...,Xn from the population X.
1 Pn
Xk . It
Let represent the population mean. Then the sample mean is
n k=1
can be shown that
2
1 Pn
=0
E lim
k=1 Xk
n n
Tosee why, notice
Pn
1 Pn
1
1
E
Xk = E ( k=1 Xk ) = n =
n k=1
n
n
2
1 Pn
Xk
E
n k=1
P
1 Pn
1
= V ar
= 2 V ar ( nk=1 Xk )
k=1 Xk
n
n
V ar (X)
1
= 2 nV ar (X) =
n
n
2
1 Pn
E lim
X
=0
k
k=1
n n
So the sample mean approaches the population mean in mean square.
219
Definition 20.3.2.
Let a = t0 < t1 < t2 < ... < tn = b represent a partition of [a, b]. Define
P
random variable In = n1
k=0 g (tk ) [Z (tk+1 ) Z (tk )], where g (tk ) is a simple
or a complex process and g (tk ) and Z (tk+1 ) Z (tk ) are independent. If
Rb 2
E g (t) dt < and the mean squared dierence between In and U is zero
a
as n :
2
E lim (In U ) = 0
n
(20.17)
Then we say
Rb
a
g (t) dZ (t) = U
In converges to
Rb
a
(i 1)2 h2 + (ih)2
2
220
RT
RT
2
2
[dZ (t)] = 0 g (t) [dZ (t)] where g (t) = 1
Partition [0, T ] into [0, h], [h, 2h],...,[(n 1) h, nh = T ].
P
Pn
2
2
Let In = n1
k=0 {Z [(k + 1) h] [Z (kh)]} =
k=1 {Z (kh) [Z (k 1) h]}
Z (kh) [Z (k 1) h] is a normal random variable with mean 0 and variance
h. Using Equation 20.9, we have
2
E{Z (kh)
Pn [Z (k 1) h]} = h
In = k=1 {Z (kh) [Z (k 1) h]}2
RT
2
Since E (In ) = nh = T , we guess that 0 [dZ (t)] = T
The dicult part is to verify that the mean square error is zero as n :
hP
i2
2
lim E ( nk=1 {Z (kh) [Z (k 1) h]}) T = 0
0
Let k = Z (kh) [Z (k 1) h]
Pn
2
( k=1 {Z (kh) [Z (k 1) h]}) = 21 + 22 + ... + 2n
hP
i2
2
n
( k=1 {Z (kh) [Z (k 1) h]}) T
2
2
2
2
1 + 22 + ... + 2n = 41 + 42 + ...4n + 221 22 + 221 23 + ...
E 4k = 3h2
E 41 + 42 + ...4n = 3nh2
i and j where i 6= j are two independent normal random variables (Point
3 of the Brownian
motion
definition)
E 2i 2j = E 2i E 2j = h h = h2
1
There are n (n 1) pairs of i and j where i 6= j
2
1
2E 21 22 + 21 23 + ... = 2 n (n 1) h2 = n (n 1) h2
2
2
E 21 + 22 + ... + 2n = 3nh2 + n (n 1) h2
T
n
2
T
3n + n (n 1) 2
[3n + n (n 1)] h T = [3n + n (n 1)]
T 2 =
T
n
n2
2
T2
As n ,
3n + n (n 1)
1
n2
3n + n (n 1) 2
T T2 0
n2
221
hP
i2
2
lim E ( nk=1 {Z (kh) [Z (k 1) h]}) T = 0
n
RT
Hence 0 [dZ (t)]2 = T .
RT
Example 20.3.4. Calculate 0 Z (t) dZ (t)
2
(a + b) a2 + b2
Use the formula: ab =
2
Let a = Z [(k 1) h]
b = Z (kh) Z [(k 1) h]
a
P+n b = Z (kh)
k=1 Z [(k 1) h] {Z (kh) Z [(k 1) h]}
2
2
P
[Z (kh)] Z [(k 1) h] {Z (kh) Z [(k 1) h]}2
= nk=1
2
1 Pn
1 Pn
2
2
{[Z
(kh)]
Z
[(k
1)
h]
}
{Z (kh) Z [(k 1) h]}2
=
k=1
2
2 k=1
1
1 Pn
2
= [Z (nh)]
{Z (kh) Z [(k 1) h]}2
2
2 k=1
1
1 Pn
2
= [Z (T )]
{Z (kh) Z [(k 1) h]}2
2
2 k=1
In =
1
1 Pn
2
{Z (kh) Z [(k 1) h]}2
[Z (T )]
2
2 k=1
Pn
lim
lim In
k=1 Z [(k 1) h] {Z (kh) Z [(k 1) h]} = n
P
1
1
2
n
= [Z (T )] lim
{Z (kh) Z [(k 1) h]}2
2
2 n k=1
Pn
From the previous example, we know that k=1 {Z (kh)Z [(k 1) h]}2 apRT
RT
proaches 0 [dZ (t)]2 = T in the mean square. So we guess that 0 Z (t) dZ (t) =
1
1
[Z (T )]2 T
2
2
2
1
1
2
Next, we need to prove that E lim In
=0
[Z (T )] T
n
2
2
1
1
1
1 Pn
2
In
{Z (kh) Z [(k 1) h]}2
[Z (T )] T = T
2
2
2
2 k=1
In E (In )
1
1 Pn
1
1
2
2
= [Z (T )]2
{Z
(kh)
Z
[(k
1)
h]}
[Z
(T
)]
T
2
2 k=1
2
2
Pn
1
2
T k=1 {Z (kh) Z [(k 1) h]}
=
2
From the previous example, we found that
2
Pn
lim E T k=1 {Z (kh) Z [(k 1) h]}2 = 0
n
2
1
1
E lim In
=0
[Z (T )]2 T
n
2
2
n
222
So we have
Z (t) dZ (t) =
1
1
2
[Z (T )] T
2
2
(20.18)
RT
0
xdx =
1
Equation 20.18 has an extra term T . This is why this extra term is
2
needed. Taking expectation of Equation 20.18:
i 1
hR
1
T
2
E 0 Z (t) dZ (t) = E [Z (T )] T
2
2
i
hR
T
As to be explained later, E 0 Z (t) dZ (t) = 0. We already know that
hR
i 1
1
T
2
2
E [Z (T )] = T . Hence E 0 Z (t) dZ (t) = E [Z (T )] T = 0. The extra
2
2
1
term T is needed so the expectations of both sides of Equation 20.18 are
2
equal.
20.4
1. Lineality.
RT
0
RT
0
RT
0
g (t) dZ (t)+c2
RT
0
h (t) dZ (t)
T
E X 2 (t) dt < , then E 0 X (t) dZ (t) =
The proof is complex. However, for a simple process g (t) and h (t), Property
#1 holds due to the definition of the stochastic integral. For a simple process
g (t), Property #2 can be easily established.
RT
P
g (t) dZ (t) = n1
k=0 g (tk ) [Z (tk+1 ) Z (tk )]
0
g (tk ) is constant during each partition interval and independent of Z (tk+1 )
Z (tk ). Then
E{g (tk ) [Z (tk+1 ) Z (tk )]} = E [g (tk )] E [Z (tk+1 ) Z (tk )] = E [g (tk )] 0 =
0
RT
RT
RT
Lets apply Property 2 to 0 Z (t) dZ (t). Since 0 E Z 2 (t) dt = 0 tdt =
h
i
RT
1 2
T < , we have E 0 Z (t) dZ (t) = 0
2
223
20.5
Itos lemma
20.5.1
Multiplication rules
dZ
dt
0
dZ
dt
dt
0
0
(dt)2 = 0
(dt) doesnt contain any Brownian motion term dZ. So we need to calculate
(dt)2 according to the deterministic calculus. In the deterministic calculus,
2
2
(dt) 0 as dt 0. Hence (dt) = 0.
The textbook Derivatives Markets also gives the following formula:
0
dZ dZ = dt
(20.19)
20.5.2
Itos lemma
Suppose that a stock has an expected instant return [S (t) , t], dividend
yield S (t) , t, and instant volatility [S (t) , t] follows geometric Brownian motion:
dS (t) = dt + dZ (t)
(20.20)
Here , , and are function of the stock price S (t) and time t.
224
Let C [S (t) , t] represent the value of a call or put option. We want to find
out the change of the option value given theres a small change of the stock price
and a small change of time. Using Taylor series, we have:
C
1 2C
2C
C
1 2C
2
2
(dS)
+
(dt)
+
dS +
dt+
dSdt (20.21)
S
t
2 S 2
2 t2
St
2
2
[dS (t)] =
dt + dZ (t) = (dt) +2 dZ (t) dt+
dC [S (t) , t] =
[dZ (t)]
2
dSdt = (dt) + dZ (t) dt
[dS (t)] = dt
Now we have:
[dZ (t)] = dt
dZ (t) dt = 0
dSdt = 0
C
C
C
1 2C
C
1 2 C 2
2
(dS)
=
dt
dS +
dt +
dS
+
dt
+
S
t
2 S 2
S
t
2 S 2
(20.22)
Next, apply Equation
20.20 toEquation ??:
C
C
1 2 C 2
dC [S (t) , t] =
dt + dZ (t) +
dt
dt +
S
t
2 S 2
C
C
1 2 2 C
C
+
dZ (t)
+
dt +
S
2 S 2
t
S
C
C
1 2 2 C
C
+
dZ (t)
(20.23)
+
dt +
dC [S (t) , t] =
2
S
2 S
t
S
dC [S (t) , t] =
[S (t) , t] = S (t)
[S (t) , t] = S (t)
[S (t) , t] = S (t)
Equation 20.23 becomes:
C
C
1
C
2C
dC [S (t) , t] = ( ) S
+ 2S 2 2 +
dt +
SdZ (t) (20.24)
S
2
S
t
S
225
Tip 20.5.2. Dont bother memorizing Equation 20.24. Just derive Equation
20.24 on the spot.
Please note that
C
C
2C
=
=
=
S
S 2
t
Then Equation 20.22 becomes:
(option Greeks)
1
2
dC [S (t) , t] = dS + dt + (dS)
2
(20.25)
20.6
There are two minor concepts under geometric Brownian motion. SOA can
easily write a question on these concepts. So lets study them.
20.6.1
X (t) h
 {z }
deterministic component
+ X (t) Y (t) h
{z
}

random comp onent
One phrase you need to know is called "the ratio of the standard deviation
to the drift" (the drift is actually the deterministic component). The ratio of
the standard deviation to the drift is defined as:
X (t) h
(20.27)
=
X (t) h
h
226
20.6.2
Let W1 (t) and W2 (t) represent two independent Brownian motions. Suppose
Z (t) = W1 (t)
0
Z (t) = W1 (t) +
p
1 2 W2 (t)
(20.28)
(20.29)
Please note that Z (t) is normally distributed with mean 0 and variance t
because W1 (t) is normally distributed with mean 0 and variance
p t.
You might wonder why Equation 20.29 has constants and 1 2 . These
0
two constants are needed to make Z (t) normally distributed with variance t.
Because W1 (t) and W2 (t) areptwo independent normal random variables, the
2
linearhcombination
i
hW1 (t) + p1 W2 (t) iis also normally distributed.
0
E Z (t) = E W1 (t) + 1 2 W2 (t)
hp
i
= E [W1 (t)] + E
1 2 W2 (t)
p
= E [W1 (t)] + 1 2 E [W2 (t)]
p
= 0 + 1 2 0 = 0
h 0 i
h
i
p
V ar Z (t) = V ar W1 (t) + 1 2 W2 (t)
hp
i
= V ar [W1 (t)] + V ar
1 2 W2 (t)
= 2 t + 1 2 t = t
h
i
0
0
The covariance between Z (t) and Z (t) is: Cov Z (t) , Z (t)
Using the
formula:
Yi) = E (XY ) hE (X)iE (Y )
h standard
i
h Cov (X,
0
0
0
Cov Z (t) , Z (t) = E Z (t) Z (t) E [Z (t)] E Z (t)
h
i
h
i
0
0
= E Z (t) Z (t) 0 0 = E Z (t) Z (t)
h
i
p
p
0
Z (t) Z (t) = W1 (t) W1 (t) + 1 2 W2 (t) = [W1 (t)]2 + 1 2 W1 (t) W2 (t)
h
i
hp
0
E Z (t) Z (t) = E [W1 (t)]2 + E
1 2 W1 (t) W2 (t)
p
2
= E [W1 (t)] + 1 2 E [W1 (t) W2 (t)]
E [W1 (t)]2 = t
E [W1 (t) W2 (t)] = E [W1 (t)] E [W2 (t)] = 0 0 = 0
i
h
0
(20.30)
E Z (t) Z (t) = t
227
h
i
h
i
0
0
The textbook says calls Cov Z (t) , Z (t) = E Z (t) Z (t) = t the corre0
lation between Z (t) and Z (t). However, the term correlation between X and
Y typically means the following:
Cov (X, Y )
X,Y =
X Y
0
If wehuse the typical
i definition, the correlation between Z (t) and Z (t) is:
0
t
= =
Z(t) Z 0 (t)
t t
Since Derivatives Markets is the textbook, we have to adopt its definition
0
that the correlation between Z (t) and Z (t) is t.
0
dZ (t) and dZ (t) are both normal random variables with mean 0 and variance dt. Applying Equation 20.30 and replacing t with dt, we have:
i
h
0
(20.31)
E dZ (t) dZ (t) = dt
0
Finally,
lets explain
why Equation 20.19 dZ dZ = dt holds.
i
h
0
E dZ (t) dZ (t) = dt
i2
h
i
h
i2
i
h
h
0
0
0
0
E dZ (t) dZ (t) dt = V ar dZ (t) dZ (t) = E dZ (t) dZ (t) E 2 dZ (t) dZ (t)
0
dZ (t) dZ (t) h
i
p
= dW1 (t) d W1 (t) + 1 2 W2 (t)
h
i
p
= dW1 (t) dW1 (t) + 1 2 dW2 (t)
p
= [dW1 (t)]2 + 1 2 dW1 (t) dW2 (t)
h
i2
0
dZ (t) dZ (t)
p
4
2
2
3
= 2 [dW1 (t)] + 1 2 [dW1 (t)] [dW2 (t)] +2 1 2 [dW1 (t)] dW2 (t)
h
i2
0
E dZ (t) dZ (t)
= 2 E [dW1 (t)]4 + 1 2 E [dW1 (t)]2 [dW2 (t)]2 +2 1 2 E [dW1 (t)]3 dW2 (t)
p
= 2 E [dW1 (t)]4 + 1 2 E [dW1 (t)]2 E [dW2 (t)]2 +2 1 2 E [dW1 (t)]3 E [dW2 (t)]
p
= 2 3 (dt)2 + 1 2 dt dt + 2 1 2 0 0
2
= 2 3 (dt) + 1 2 dt dt
= 32 + 1 2 (dt)2 = 22 + 1 (dt)2
h
i
0
E dZ (t) dZ (t) = dt
i2
i
h
h
0
0
E dZ (t) dZ (t) E 2 dZ (t) dZ (t)
2
2
2
= 22 + 1 (dt) 2 (dt) = 2 + 1 (dt)
228
(dt) = 0
h
i2
i
h
0
0
E dZ (t) dZ (t) E 2 dZ (t) dZ (t) = 0
h
i2
h
i
0
0
E dZ (t) dZ (t) dt = V ar dZ (t) dZ (t) = 0
0
dZ dZ = dt
(20.32)
2
2
2
2
2
(dS) = S (dt) + 2dtdZ + 2 (dZ) = S 2 2 0 + 2 0 + 2 dt =
S 2 2 dt
1 2 ln S
2
(dS) .
2 S 2
1 2
1 2
dS
1 2
d ln S =
dt = dt + dZ dt = dt + dZ
S
2
2
2
During a tiny time interval [t, t + dt], the change of ln S is d ln S.
d ln S is a normal random variable. This is why.
229
During the fixed interval [t, t + dt], dt is a constant. Since a constant plus a
random variable is also a random variable,
1
1
d ln S = 2 dt+dZ is a normal random variable with mean 2 dt
2
2
and variance 2 dt
Now consider the time interval [0, t].
Rt
Rt
Rt
1 2
ln S (t) = ln S (0) + 0 d ln S = ln S (0) + 0 ds + 0 dZ
2
Rt
1 2 Rt
= ln S (0) +
ds + 0 dZ = ln S (0) + 0.5 2 t + Z
0
2
Z is a standard
normal random variable with mean 0 and variance 1, that
2
E [S (t)] = S (0) e(0.5 )t E eZ
(DM 18.13)
2
E eZ = e0+0.5 t
h
i
2
2
E [S (t)] = S (0) e(0.5 )t e0+0.5 = S (0) et
230
I want you to memorize the following results (these results are used over and
over in Exam MFE):
dZ N (0, dt)
(20.33)
Z N (0, t)
(20.34)
x N m, v 2
E (ex ) = em+0.5v
(20.35)
(20.36)
(20.37)
dS (t)
= dt + dZ (t)
S (t)
2
S (t) = S (0) e(0.5 )t+Z
dS (t)
= dt + dZ (t)
S (t)
20.7
dS (t)
= dt + dZ (t)
S (t)
d ln S (t) = 0.5 2 dt + dZ
dS (t)
= dt + dZ (t)
S (t)
dS (t)
= dt+dZ (t)
S (t)
E [S (t)] = S (0) et
(20.39)
(20.40)
Sharpe ratio
If two nondividend paying assets are perfectly corrected (i.e. they are driven
by the same Brownian motion Z (t)), then their Sharpe ratios are equal. Lets
derive this.
The price processes of Asset 1 and Asset 2 are:
SR =
dS1 = 1 S1 dt + 1 S1 dZ
(20.42)
dS2 = 2 S2 dt + 2 S2 dZ
(20.43)
231
1
1 S1
1
2 S2
dS1 =
1
dt + dZ
1
(20.44)
dS2 =
2
dt + dZ
2
(20.45)
1
units of Asset 1 and short sell
1 S1
1
units of Asset 2. If we hold one unit of Asset 1 at time zero, then
2 S2
after a tiny interval dt, the value of Asset 1 increases by the amount dS1 . If
1
units of Asset 1 at time zero, then after dt the value of
we hold N1 =
1 S1
1
1
1
N1 =
units of Asset 1 will increase by
dt + dZ. Notice
dS1 =
1 S1
1 S1
1
that the increase of the value of Asset 1 has a random component dZ, where dZ
is a normal random variable with mean 0 and variance dt.
1
Similarly, if we short sell N2 =
units of Asset 2 at time zero, after dt,
2 S2
1
2
the value of Asset 2 will increase by
dt + dZ. The increase of
dS2 =
2 S2
2
the value of Asset 2 has a random component dZ, where dZ is a normal random
variable with mean 0 and variance dt.
Suppose at time zero we simultaneously buy N1 units of Asset 1 and short
sell N2 units of Asset 2. Then at dt, we close our position by selling N1 units
of Asset 1 in the open market and buying N2 units of Asset 2 from the open
market.
The cash flow at time zero:
1
1
dollar to buy N1 units of Asset 1.
S1 =
We pay N1 S1 =
1 S1
1
N2 =
We receive N2 S2 =
1
dollars for short selling N2 units of Asset 2
2
1
1
1
1
dollars. If
1
1
1 2
1
1
1
1
1
1
and borrow
dollars. If
.
1 2
1 2
1 2
Our net cash outgo is zero.
Our payo at time dt
At time dt, we sell o N1 units of Asset 1 in the open market for the price
1
1
of S1 + dS1 , receiving N1 (S1 + dS1 ) = N1 S1 + N1 dS1 =
+
dt + dZ
1
1
232
erdt
1 2
Our profit at dt is:
P rof it =
1
1
1
2
1
1
+
dt + dZ
+
dt + dZ
erdt (20.46)
1
1
2
2
1
2
1
1
1
1
1
1
2
1
1
1
+
dt
+
dt
erdt
1
1 2
2
1
2
1
1
1
1 2
1
=
+
dt
erdt
2
1 2 1
1 2
1
1
1
1
1 2
1
1
=
+
dt
rdt
1
2
1 2
1
2
1 2
1 2
1
1
=
dt
rdt
2
1
1 2
1 r 2 r
=
dt
1
2
= (SR1 SR2 ) dt
1 r 2 r
P rof it =
dt = (SR1 SR2 ) dt
1
2
(20.47)
Equation 20.46 and 20.47 dont have the random term dZ (t), indicating that
the profit is surely made.
Our cash outgo is zero at t = 0, but well have the profit indicated by 20.47
at dt.
1 r 2 r
Wealth 0
dt = (SR1 SR2 ) dt
1
2
Time
0
dt
To avoid arbitrage, the profit needs to be zero:
1 r
2 r
=
SR1 = SR2
1
2
20.8
233
This section is dicult because the author tired to put too many complex concepts in this small section. It seems that the author was in a big hurry to finish
this chapter. The author mentioned many concepts (such as martingale, Girsanovs theorem) but he didnt really explain them, leaving us hanging in the
air asking why.
The only thing worth studying is how to transform a standard geometric
Brownian motion into a risk neutral process.
The standard geometric Brownian motion is:
dS (t)
= ( ) dt + dZ (t)
S (t)
This is how to transform:
dS (t)
= ( ) dt + dZ (t)
S (t)
= (r ) dt + dZ
(t) + ( ) dt
dt
= (r ) dt + dZ (t) +
dt
Define dZ (t) = dZ (t) +
dS (t)
= ( ) dt + dZ (t) = (r ) dt + dZ (t)
S (t)
Just learn this transformation and move on.
20.9
Valuing a claim on S a
20.9.1
Process followed by S a
234
S A
1
1 2S A
2
2
(dS) = AS A1 dS + A (A 1) S A2 (dS)
dS +
S
2 S 2
2
2
2
However, (dS) = [( ) dt + dZ] S 2 = 2 dt S 2 = 2 S 2 dt
dS A =
1
dS A = AS A1 dS + A (A 1) S A2 2 S 2 dt
2
1
= AS A1 dS + A (A 1) S A 2 dt
2
dS A
dS
1
=A
+ A (A 1) 2 dt
SA
S
2
2
=A
( ) dt + AdZ + 0.5A2(A
1) dt
= A ( ) + 0.5A (A 1) dt + AdZ
20.9.2
2
E S A (t) = S A (0) e[A()+0.5A(A1) ]t
2
We can also derive E S A (t) using S A (t) = S A (0) e[A()0.5A ]t+AZ
2
E S A (t) = E S A (0) e[A()0.5A ]t+AZ
2
= S A (0) E e[A()0.5A ]t+AZ
2
= S A (0) E e[A()0.5A ]t eAZ
2
= S A (0) e[A()0.5A ]t E eAZ
2 2
E eAZ = e0.5A t
A
2
2
2 2
E S (t) = S A (0) e[A()0.5A ]t e0.5A t = S A (0) e[A()+0.5A(A1) ]t
235
A
2
E S (t) = E S A (0) eA(0.5 )t+AZ
2
= S A (0) eA(0.5 )t E eAZ
2
2 2
= S A (0) e[A()0.5A ]t e0.5A t
2
= S A (0) e[A()+0.5A(A1) ]t
20.9.3
dS A
= A ( ) + 0.5A (A 1) 2 dt + AdZ
A
S
= ( ) dt + AdZ
where is the expected return on a claim on S A (t) and is the continuous
dividend yield earned by S A (t). The textbook calls the lease rate.
These two processes have the same risk dZ. Consequently, they have the
same Sharpe ratio:
r
r
=
= r + A ( r)
A
where r is the continuously compounded riskfree interest rate.
20.9.4
Specific examples
The textbook keeps mentioning Jensens inequality. So lets first talk about
Jensens inequality. Jensens inequality is in the appendix C of the textbook.
You can also find information at http://en.wikipedia.org/wiki/Convex_
function
Jensens inequality says
1. if f (x) is convex, then for any probability distribution, we have E [f (x)]
f [E (x)]
2. if f (x) is concave, then for any probability distribution, we have E [f (x)]
f [E (x)]
236
If at any point you draw a tangent line, the function f (x) stays above the
tangent line, then f (x) is a convex function.
If at any point you draw a tangent line, the function f (x) stays below the
tangent line, then f (x) is a concave function.
237
d2 y
= 2 > 0, y = x2 is a convex function. Suppose
dx2
wetake two points A (1, 1) and
B (4, 16). The mid point of the line AB is
1+4
1 + 16
C
= 2.5,
= 8.5 . The fact that y = x2 is a convex function means
2
2
that y = x2 curves up. Then it follows that the point C must be above the point
12 + 42
D 2.5, 2.52 = 6. 25 . From the graph below, we clearly sees that
= 8.5
2
2
1+4
= 6.25 (which
(which is the height of the point C) is greater than
2
is the height of the point D). This is an example where the mean of a convex
function is greater than the function of the average.
Consider y = x2 . Sine
25
20
B
15
10
C
D
A
0
0
12 + 42
>
2
1+4
2
238
1+4
1+ 4
and B (4, 2). The mid point of AB is C
= 2.5,
= 1.5 . Since
2
2
1+4
y = x curves down, then it follows that C must be lower than D
= 2.5, 2.5 = 1. 58 .
2
1+ 4
From the graph below, we clearly sees that
= 1.5 (which is the height
2
r
1+4
= 1. 58 (which is the height of the point
of the point C) is less than
2
D). This is an example where the mean of a concave function is less than the
function of the average.
2.0
D
C
1.5
1.0
0.5
0.0
0
1+ 4
1+4
<
, an example of E [f (x)] f [E (x)]
2
2
By now you should have intuitive feel of Jensens inequality. Lets move on.
239
2
2
P
S (T ) = erT S 2 (0) e[2(r)+ ]T
V (0) = F0,T
As a general rule, the forward price of any asset at T is just the prepaid
forward price accumulating at the riskrate from time 0 to T :
P
erT
F0,T = F0,T
We use the riskfree rate r in the above equation. To get an asset at T , we
P
as a buyer can either pay the seller F0,T
at time 0 or pay the seller F0,T at time
T . To avoid the arbitrage, the two payments should dier only in timing. So
P
F0,T = F0,T
erT .
2
2
2
P
S (T ) erT = S 2 (0) e[2(r)+ ]T
F0,T S (T ) = F0,T
P
F0,T [S (T )] = F0,T
[S (T )] erT = S (0) eT erT = S (0) e(r)T
2
2
2
F0,T S (T ) = S 2 (0) e[2(r)+ ]T = (F0,T [S (T )])2 e T
2
Since e T 1, we have F0,T S 2 (T ) (F0,T [S (T )])2 . This agrees with
Jensens inequality. Roughly speaking 1 , F0,T = E (ST ).
d2 2
S = 2 > 0. Hence S 2 is convex.
dS
2
According to Jensens inequality, we have E S 2 (T ) (E [S (T )]) . This
2
2
2
leads to F0,T S (T ) = E S (T ) (F0,T [S (T )]) = (E [S (T )]) .
S 2 is twice dierentiable and
1
at T is:
If A = 1, the time 0 value of the claim
S (T )
1
1 [(r)+2 ]T
1
2
P
= erT
e T
V (0) = F0,T
= erT
e
(r)T
S
(T
)
S
(0)
S
(0)
e
1
1
1
2
P
F0,T
e T
= F0,T
erT =
S (T )
S (T )
S (0) e(r)T
(r)T
However,
F0,T [S
(T )] = S (0) e
1
1
1
2
F0,T
=
e T
S (T )
F0,T [S (T )]
F0,T [S (T )]
240
1
Since
is concave, according to Jensens inequality, we have:
S (T )
2
2
1
1
1
1
= E
.
F0,T
=E
F0,T
S (T )
S (T )
S (T )
S (T )
d2 1
2
S = 3 > 0. Hence S 2 is convex.
2
dS
S
2
1
1
According to Jensens inequality, we have E
. This
E
S (T )
S (T )
2
2
1
1
1
1
leads to F0,T
= E
.
=E
F0,T
S (T )
S (T )
S (T )
pS (T )
If A = 0.5, the
the claim S (T ) at T is:
hptime 0ivalue of p
2
P
rT
S (T ) = e
S (0)e[0.5(r)+0.50.5(0.51) ]T
V (0) = F0,T
p
2
= herT S i(0)e[0.5(r)0.125 ]T
p
p
p
2
F0,T
S (T ) = S (0)e[0.5(r)0.125 ]T = S (0)e0.5(r)T
p
p
2
2 T
= S (0) e(r)T e0.125
=
F0,T [S (T )]e0.125 T
i
h
p
p
2
Since e0.125 T 1, we see that F0,T
S (T ) F0,T [S (T )]
This agrees with Jensens inequality.
p
p
d2
S (T ) is twice dierentiable.
S = 4S 1.5 < 0. Hence S (T ) is
dS 2
concave. hp
i
hp
i p
hp
i
S (T ) = E
S (T ) F0,T [S (T )] =
E (S (T ))
F0,T
S 1 is twice dierentiable and
Example 20.9.1.
The price of a stock follows a geometric Brownian motion:
dS (t)
= (0.1 0.04) dt + 0.3dZ
S (t)
The current price of the stock is 10.
The continuously compounded dividend yield is 0.04 per year.
The continuously compounded riskfree rate is 0.06 per year.
The seller and the buyer enter a forward contract. The contract requires the
seller to pay the buyer S 2 (5) five years from now.
Calculate
the prepaid forward price
the forward price
241
A
2
P
S (T ) = erT S A (0) e[A(r)+0.5A(A1) ]T
F0,T
2
2
P
F0,5
S (5) = e0.065 102 e(2(0.060.04)+0.52(21)0.3 )5 = 141. 906 8
P
F0,5 S (5) = F0,5
S 2 (5) e0.065 = 141. 906 8e0.065 = 191. 554 1
= r + A ( r) = 0.06 + 2 (0.1 0.06)
= 0.14
= A ( ) + 0.5A (A 1) 2
2
2
2
2
2
ln
S (5) N ln 10 + 2 (0.1 2 0.04) 0.5 2 0.3 5, 2 0.3 5
2 (0.1 0.04) 0.5 2 0.3 5 = 0.15
(z) = 0.594 3
P S 2 (5) 160 = 0.594 3
Example 20.9.2.
242
The probability that
1
0.03.
S (4)
Solution.
dS (t)
= (0.12 0.05) dt + 0.25dZ
S (t)
2
P
F0,T
S A (T ) = erT S A (0) e[A(r)+0.5A(A1) ]T
2
1
1
P
F0,4
= e0.074 e(1(0.070.05)+0.5(1)(11)0.25 )4 = 0.04479
S (4)
20
1
1
P
F0,4
= F0,4
e0.074 = 0.04479e0.074 = 0.05 926
S (4)
S (4)
= r + A ( r) = 0.07 1 (0.12 0.07)
= 0.02
= A ( ) + 0.5A (A 1) 2
1
1
2
ln
N ln
+ 1 (0.12 0.05) 0.5 (1) 0.252 4, (1) 0.252 4
20
S (4)
1
1
ln + 1 (0.12 0.05) 0.5 (1) 0.252 4 = ln 0.155 = 3. 150 7
20
20
1
1
0.03 = P ln
ln 0.03 = (z)
P
S (4)
S (4)
ln 0.03 (3. 150 7)
= 0.711 7
z=q
2
(1) 0.252 4
1
(z) = 0.238 3
P
0.03 = 0.238 3
S (4)
Example 20.9.3.
The price of a stock follows a geometric Brownian motion:
dS (t)
= (0.15 0.04) dt + 0.35dZ
S (t)
The current price of the stock is 10.
The continuously compounded dividend yield is 0.04 per year.
The continuously compounded riskfree rate is 0.08 per year.
The seller and the buyer
p enter a forward contract. The contract requires the
seller to pay the buyer S (6) six years from now.
Calculate
243
p
S (6) 4.
Solution.
dS (t)
= (0.15 0.04) dt + 0.35dZ
S (t)
2
P
F0,T
S A (T ) = erT S A (0) e[A(r)+0.5A(A1) ]T
i
hp
2
P
F0,6
S (6) = e0.086 10e(0.5(0.080.04)+0.50.5(0.51)0.35 )6 = 2. 012 6
i
i
hp
hp
P
F0,6
S (6) = F0,6
S (6) e0.086 = 2. 012 6e0.086 = 3. 252 5
= r + A ( r) = 0.08 + 0.5 (0.15 0.08) = 0.115
= A ( ) + 0.5A (A 1) 2
ln
S (6) N ln 10 + 0.5 (0.15 0.04) 0.5 0.5 0.352 5, 0.52 0.352 6
244
Chapter 21
BlackScholes equation
21.1
21.1.1
Valuation equation
Lets consider a riskfree world where all assets just earn the risk free interest
rate. Support at time t we spend S (t) to buy one share of a stock. Then at t+h,
we receive D (t + h) h amount of the dividend, where D (t + h) represents the
dividend accumulated per unit of time during [t, t + h]. At t + h after receiving
D (t + h) h dividend, we sell the stock and receive S (t + h). The total amount
of money we have at t + h is D (t + h) h + S (t + h). Had we put S (t) amount of
money in a savings account, we would have S (t) (1 + rh ) amount of money at
t+h, where rh represent the (notannualized) riskfree interest rate per h period.
To avoid arbitrage, we need to have S (t) (1 + rh ) = D (t + h) h + S (t + h).
Rearranging this equation, we get:
D (t + h) h + S (t + h)
1 + rh
(DM 21.1)
S (t + h) S (t) + D (t + h) h = rh S (t)
(DM 21.2)
S (t) =
hr
i
S (t + h) S (t)
dS (t)
h
= lim
= lim
S (t) D (t + h) = rS (t) D (t)
h0
h0 h
dt
h
(DM 21.3)
1
In Equation 21.3, r = rh represents the annualized continuously comh
pounded interest rate per year. rh is the interest rate per h period; the total
1
1
number of h lengths in a year is . Hence r = rh represents the annualized
h
h
continuously compounded interest rate per year.
245
246
21.1.2
Bonds
dS (t)
= rS (t).
dt
(DM 21.4)
21.1.3
At time t, you spend S (t) and buy one share of a stock. This gives you two
things:
at time T you can sell the stock and get S (T ), which is worth S (T ) er(T t)
at t
you accumulate dividend at a continuous rate of D (s), where D (s) is the
instant dividend earned per unit of time at time s. The total value of
the continuous dividend earned during the interval [s, s + ds] is D (s) ds,
which is worth [D (s) ds] er(st) = D (s) er(st) ds at time t. The present
value at time t of the total continuous dividend earned during the interval
RT
[t, T ] is t D (s) er(st) ds The PV of this continuous flow of dividend is
1 er(T t)
DaT tr = D
r
RT
To avoid arbitrage, we have S (t) = S (T ) er(T t) + t D (s) er(st) ds.
RT
Please note that if D (s) = D is a constant, then t D (s) er(st) ds =
RT
1 er(T t)
D t er(st) ds = DaT tr = D
.
r
RT
Anyway, t D (s) er(st) ds is a continuous annuity. If you have trouble
RT
understanding t D (s) er(st) ds, refer to your FM book.
21.2
BlackScholes equation
21.2.1
(DM 21.11)
247
(DM 21.7)
(DM 21.8)
dI is the interest earned on the option during [t, t + dt]. Since W is invested
in a savings account, we have dW = rW dt. This says that the interest earned
on W during [t, t + dt] is rW dt. Notice that the change of S is dS + Sdt (the
sum of the change of the stock price dS and the dividend received Sdt). Apply
Ito lemma:
dV = Vt dt + VS dS + 0.5 2 S 2 VSS dt
dI = Vt dt + VS dS + 0.52 S 2 VSS dt + N (dS + Sdt) + dW = Vt dt +
(VS + N ) dS + 0.5 2 S 2 VSS dt + N Sdt + dW
Set N = VS . Then dS term becomes zero and W = (V VS S).
Because our initial cost is zero, the interest we earned dI should be zero.
dI = Vt dt + 0.5 2 S 2 VSS dt VS Sdt r (V VS S) dt = 0
This leads to DM 21.11.
Vt + 0.52 S 2 VSS + (r ) SVS rV = 0
21.2.2
Simple PV calculation
Verification that the price of a zerocoupon bond satisfies the BlackScholes
equation DM 21.11.
$1 at time T is worth V (t) = er(T t) at t.
=
VS = VSS = 0
Vt = rer(T t) = rV
2 2
=
Vt + 0.5 S VSS + (r ) SVS rV = 0
Verification that the price of a prepaid forward contract satisfies the BlackScholes equation DM 21.11.
V (S, t) = S (t) e(T t)
=
VS = e(T t)
VSS = 0
Vt = S (t) e(T t)
=
Vt +0.5 2 S 2 VSS +(r ) SVS rV = Se(T t) +(r ) Se(T t)
rSe(T t) = 0
248
Call option
The textbook explains that the price of a European call option satisfies (1) the
boundary condition and, (2) DM 21.11.
V = S (t) e(T t) N (d1 ) Ker(T t) N (d2 )
Verification that the call price formula meets the boundary condition.
The boundary condition is that at the call expiration date T the call is worth
S (T ) K (T ) if S (T ) > K (T ) and zero otherwise.
1
S (t)
+ r + 2 (T t)
K
2
d1 =
T t
d2 = d1 T t
S (T )
1 2
ln
+ r + (T t)
K
2
, d2 = d1 ,
If t approaches T , then d1 =
T t
and
1 2
S (T )
r+
T t
ln
S (T )
S (T )
2
If S (T ) > K, then
> 1, ln
> 0, d1 = K +
=
K
K
T t
+ , N (d1 ) = N (d2 ) = 1, V = S T.
ln
S (T )
If S (T ) < K, then ln
< 0, d1 = , N (d1 ) = N (d2 ) = 0 and
K
V =0
By the way, we dont need to worry about S (T ) = K because the probability of S (T ) = K is zero. The probability that a continuous random variable
takes on a fixed value is zero. When we talk about the probability regarding a
continuous random variable X, we talk about the probability that X falls in a
range [a, b], not the probability that X takes on a single value. If the probability
that X takes a single value is not zero, then the total probability that a < X < b
will be infinite because there are infinite number of single values in the range
[a, b].
We have proved that the call price satisfies the boundary condition.
For the verification that the call price satisfies DM 21.11, see my solution to
DM Problem 21.5, 21.6, and 21.7.
We can also verify that the European put price satisfies DM 21.11. The put
price is
V = Ker(T t) N (d2 ) S (t) e(T t) N (d1 )
Using the formula N (x) = 1 N (x), we can rewrite the put price as
V = Ker(T t) [1 N (d2 )] S (t) e(T t) [1 N (d1 )] = Ker(T t)
S (t) e(T t) + S (t) e(T t) N (d1 ) Ker(T t) N (d2 )
249
You can also derive the above equation using the putcall parity. Notice that
S (t) e(T t) N (d1 ) Ker(T t) N (d2 ) is the call price.
Each of the three terms, Ker(T t) , S (t) e(T t) , and S (t) e(T t) N (d1 )
r(T t)
Ke
N (d2 ), satisfies the BS PDE. Hence the put price satisfies the BS
PDE.
I wont prove that the put price satisfies the boundary condition V (t = T ) =
K S (T ) if K > S (T ) and zero otherwise. You can easily prove this yourself.
Key formula to remember:
If t T , then d1 = d2 and
If S (T ) > K, N (d1 ) = N (d2 ) = 1
If S (T ) < K, N (d1 ) = N (d2 ) = 0
All or nothing option
The textbook points out that S (t) e(T t) N (d1 ) and er(T t) N (d2 ) each satisfy the BS PDE (see my solution to DM Problem 21.5, 21.6). So S (t) e(T t) N (d1 )
can be a price of a derivative; er(T t) N (d2 ) can be a price of another derivative. What derivatives are priced as S (t) e(T t) N (d1 ) and er(T t) N (d2 )
respectively?
Lets check the boundary condition. As t T ,
S (t) e(T t) N (d1 ) S (T ) if S (T ) > K
S (t) e(T t) N (d1 ) 0 if S (T ) < K
S (t) e(T t) N (d1 ) must be the price of an option that pays S (T ) if S (T ) >
K and zero otherwise. Such an option is an assetornothing option.
Similarly, as t T
er(T t) N (d2 ) 1 if S (T ) > K
er(T t) N (d2 ) 0 if S (T ) < K
er(T t) N (d2 ) must be the price of an option that pays 1 if S (T ) > K and
zero otherwise. Such an option is an cashornothing option.
We can break down a European call option into one assetornothing option
and several cashornothing options. Buying a European call option is equivalent
to buying one assetornothing option and selling K units of cashornothing
option (you can verify that they have the same payo). Hence the price of the
European call option is S (t) e(T t) N (d1 ) Ker(T t) N (d2 ).
Similarly, we can break down a gap option (gap option is explained in Chapter 14). In a gap call, the payo is S (T ) K1 if S (T ) > K2 . This is equivalent
250
to buying one assetornothing call and selling K1 units of cashornothing option. So the price
of this gap call is S (t) e(T t) N (d1 ) K1 er(T t) N (d2 ),
(T
t)
and d2 = d1 T t.
where d1 =
T t
P (ST > K) = N (d2 ). This is the riskneutral probability that the call
will be exercised (i.e. call will finish in the money)
P (ST < K) = 1 N (d2 ) = N (d2 ). This is the riskneutral probability
that the call will NOT be exercised.
Ker(T t) N (d2 ) is the strike price multiplied by the riskneutral probability that the strike price will ever be paid. This is the expected value of
the strike price *if* the call will be exercised
S (t) e(T t) N (d1 ) is the expected value of the stock price *if* the call
will be exercised (i.e. if ST > K)
N (d1 ) is the expected fractional share of the stock *if* the call will be
exercised
If an option pays one stock when ST > K and zero otherwise, then this
option is worth S (t) e(T t) N (d1 )
If an option pays $1 when ST > K and zero otherwise, then this option is
worth PV of $1 (which is er(T t) ) multiplied by P (ST > K)
21.2.3
This section derives the BS PDE using the idea that the actual expected return
on an option should be equal to the equilibrium expected return.
The expected continuously compounded return on an option is option =
E (dV )
. Here dV is the increase of the option value (i.e. the interest earned
V dt
on the option) during [t, t + dt]. At time t, if you spend V dollars and buy an
option, then during the tiny interval [t, t + dt], your option value will go up by
dV
dV . The (not annualized) return earned on the option per dt period is
.
V
1
Since the number of dt periods in one year is , the annualized (continuously
dt
dV
1
dV
compounded) rate of return per $1 invested in the option is
=
.
V
dt
V
dt
E (dV )
dV
=
. Here V and dt are
The expected return on the option is E
V dt
V dt
treated as constants. dV is a random variable.
To calculate E (dV ), we use Itos lemma:
251
=
V
V
V
(DM 21.18)
E (dV )
SVS dZ
dV
=
V
V
V
Consider two assets, an stock and an option on the stock.
dS
The process of the stock price:
= dt + dZ
(DM 20.1)
S
dV
The process of the option price:
= option dt + option dZ
V
These two processes are driven by the same dZ. According to Chapter 20,
the stock and the option on the stock must have the same Sharpe ratio. Hence
r
option r
=
option
The correct formula should be:
=
=
V
r
option r
r =
(option r)
=
SV
SVS
S
V
V
0.5 2 S 2 VSS + ( ) SVS + Vt
r =
r
SVS
V
( r) SVS = 0.5 2 S 2 VSS + ( ) SVS + Vt rV
Vt + 0.5 2 S 2 VSS + (r ) SVS rV = 0 (BS PDE)
252
SVS
By the way, according to DM 12.8, the option elasticity is =
. Hence
V
SVS
option =
= (DM 21.20). DM 21.20 is slightly dierent from DM
V
12.9:
(DM 12.9)
option = 
option =
(DM 21.20)
Obviously, the author of Derivatives Markets changed his definition of option .
In Chapter 12, option is non negative; in Chapter 21, option can be positive,
zero, or negative.
21.3
Riskneutral pricing
This section repeats the old idea of risk neutral pricing. Under risk neutral
pricing, we can set the expected return on the stock to r and get the correct
price of a derivative.
This section contains many formulas. Make sure you understand the meaning
of each formula.
The textbook lists the following equations:
dS
= (r ) dt + dZ
S
d
E (dV ) = Vt + 0.52 S 2 VSS + (r ) SVS
dt
d
E (dV ) = rV
dt
(DM 21.28)
(DM 21.30)
(DM 21.31)
Chapter 22
Exotic options: II
This is an easy chapter.
22.1
Allornothing options
Cashornothing option.
A cash call pays $1 at T is ST > K and $0 if ST K
A cash put pays $1 at T is ST < K and $0 if ST > K
Assetornothing option.
An assetornothing call pays ST at T is ST > K and $0 if ST K
An assetornothing put pays ST at T is ST < K and $0 if ST > K
Price of cashornothing option.
The price of a cash call option is er(T t) N (d2 ), where N (d2 ) = P (ST > K)
(i.e. the risk neutral probability of ST > K)
The price of a cash put option is er(T t) [1 N (d2 )] = er(T t) N (d2 ),
where N (d2 ) = P (ST < K) (i.e. the risk neutral probability of ST <
K)
Price of assetornothing option.
The price of an asset call option is St e(T t) N (d1 )
The price of an asset put option is St e(T t) [1 N (d1 )] = St e(T t) N (d1 )
253
254
and d2 = d1 T t.
Deltahedging allornothing options
Its dicult to hedge an allornothing option because the payo is not continuous. The textbook explains this well. Refer to the textbook.
Chapter 23
Volatility
The concept of implied volatility is explained in DM section 12.5. The implied
volatility of an option is the volatility implied by the market price of the option
based on the BlackScholes formula. If you plug the implied volatility into the
BlackScholes formula, the formula should produce the option price that is equal
to the current market price of the option.
The BlackScholes formula assumes that a given stock has a constant volatility. Under the BS formula, the stocks volatility is a inherent characteristic of
the stock; it doesnt depend on other factors (such as the stick price K and the
options expiry T ). In reality, however, the implied volatility of a stock depends
on K and T .
The volatility smile is a longobserved pattern where atthemoney options
tend to have lower implied volatilities than in or outofthemoney options.
The implied volatility is lowest when the option is at the money
The implied volatility gets higher and higher as the option gets more and
more inthemoney
The implied volatility gets higher and higher as the option gets more and
more outofthe money.
If we plot the implied volatility in a 2D plane (setting the implied volatility
as Y and the strike price K as X), the diagram looks like a letter U (the letter
U looks like a smile).
To see a diagram of the volatility smile, refer to
http://www.optiontradingpedia.com/volatility_smile.htm
http://en.wikipedia.org/wiki/Volatility_smile
Volatility surface is a 3D diagram of the implied volatility as a function
of the strike price K and time to maturity T .
255
256
To account for the factor that the implied volatility depends on K and T ,
dS (t)
dS (t)
we can rewrite DM 20.25
= ( ) dt + dZ (t) as DM 23.1
=
S (t)
S (t)
( ) dt+ (St , Xt, t) dZ (t). In DM 23.1, the instant volatility (St , Xt, t) dZ (t)
is a function of the stock price St , another factor Xt, , and the
timent.
1
1 P 2
2
Historical volatility. Please note that DM 23.2 =
is not
h n 1 i=1 i
new. This formula is already used in DM Chapter 11 "Estimating Volatility." In
n
1 P
2 is the estimated variance per h period. Since the number
DM 23.2,
n 1 i=1 i
n
1
1
1 P
2i .
of h periods in one year is , the variance per year is
h
h n 1 i=1
This is all you need to know about Chapter 23.
Chapter 24
24.1.1
Duration and convexity are explained in Derivatives Markets Section 7.3. Section 7.3 is excluded from the MFE syllabus. However, Derivatives Markets page
781 mentions the duration hedging:
...hedging a bond portfolio based on duration does not result in a perfect
hedge. Recall that the duration of a zerocoupon bond is the bonds time to
maturity...
Because Chapter 24 mentions duration and Chapter 24 is on the syllabus,
SOA may test your knowledge about duration hedging. So read DM Section
7.3 and take a quick review of duration and convexity. Next, lets solve a few
problems.
Example 24.1.1.
The yield to maturity of a 5year zerocoupon bond is 6%. The face amount
of the bond is 100.
Calculate
the value of the bond
the Macaulay duration
the modified duration
the convexity
257
258
P
P1 = P0 + P = P0 1 +
P0
1
1
P
= Dmod y+ C (y)2 = 4. 716 980.01+ 26. 699 893(0.01)2 =
P0
2
2
0.045 8
P1 = 74. 725 8 (1 0.045 8) = 71. 303 3
This is very close to the correct amount of 71. 298 6.
If the yield to maturity is now 5%, the bond value is:
P1 = 100 1.055 = 78. 352 6
We can also use duration and convexity to approximate the bond value under
the new yield.
P
1
1
2
= Dmod y + C (y) = 4. 716 98 (0.01) + 26. 699 893
P0
2
2
(0.01)2 = 0.048 5
P1 = 74. 725 8 (1 + 0.048 5) = 78. 350 0
This is very close to the correct amount of 78. 352 6.
Example 24.1.2.
259
The yield to maturity of a 4year zerocoupon bond is 8%. The face amount
of the bond is 100.
Calculate
the value of the bond
the Macaulay duration
the modified duration
the convexity
In addition, calculate the bond value if the yield to maturity
moves up to 9%
moves down to 7%.
Solution.
The bond value is P0 = 100 1.084 = 73. 5030
DMac = T = 4 years.
DMac
4
Dmod =
=
= 3. 703 7 years
1 + yield
1.08
45
T (T + 1)
C=
2 = 1.072 = 17. 468 8
(1 + y)
If the yield to maturity is now 9%, the bond value is:
P1 = 100 1.094 = 70. 842 5
We can also use duration and convexity to approximate the bond value under
the new yield.
The new bond value is
P
P1 = P0 + P = P0 1 +
P0
1
1
P
2
2
= Dmod y + C (y) = 3. 703 7 0.01 + 17. 468 8 (0.01) =
P0
2
2
0.036 2
P1 = 73. 5030 (1 0.036 2) = 70. 842 2
This is very close to the correct amount of 70. 842 5.
If the yield to maturity is now 7%, the bond value is:
P1 = 100 1.074 = 76. 289 5
We can also use duration and convexity to approximate the bond value under
the new yield.
P
1
1
= Dmod y + C (y)2 = 3. 703 7 (0.01) + 17. 468 8
P0
2
2
2
(0.01) = 0.037 9
P1 = 73. 5030 (1 + 0.037 9) = 76. 288 8
This is very close to the correct amount of 76. 289 5.
260
Example 24.1.3.
Portfolio A consists of one unit of 4year zero coupon bond.
Portfolio B consists of x units of 1year zero coupon bond and y units of
8year zero coupon bond. Portfolio B is used to duration hedge Portfolio A.
Each of the three zerocoupon bonds above has 100 face amount.
The current annual eective interest rate is 5% for all three bonds.
Assume that the yield curve changes by a uniform amount if theres a change.
Demonstrate why durationhedging leads to arbitrage under two scenarios:
the annual eective interest rate moves up to 6% immediately after the
three bonds are issued.
the annual eective interest rate instantly moves down to 4% immediately
after the three bonds are issued.
Solution.
To duration hedge Portfolio A, we need to satisfy two conditions:
Portfolio A and B have the same present value
Portfolio A and B have the same duration
So we set up the following equations:
100
100
100
x+
y=
1.051
1.058
1.054
100
100
x
y
8
1.051
1.05
1+
8=4
100
100
100
100
x
+
y
x
+
y
1.051
1.058
1.051
1.058
This is the meaning of the second equation. Portfolio B consists of two
bonds. If a portfolio consists of multiple bonds, then the portfolios duration is
just the weighted average of the each bonds duration, with weight equal to the
present value of each bond.
The 2nd equation can be simplified as
100
100
100
x 1+
y 8 = 4
1
8
1.05
1.05
1.054
100
100
100
x+
y=
1
8
1.05
1.054
1.05
100
100
100
x 1+
y 8=4
1.051
1.058
1.054
x = 0.493 62 , y = 0.520 93
261
100
100
100
x+
y=
= 82. 270 2
1.051
1.058
1.054
100
= 82. 270 2
1.054
100
100
100
x+
y=
= 82. 270 2
1
8
1.05
1.05
1.054
100
= 82. 270 2
1.054
262
Why is Portfolio B better than Portfolio A? It turns out B has high convexity.
Convexity of A:
TA (TA + 1)
45
=
= 18. 1406
CA =
2
2
1.05
(1 + y)
Convexity of B is just the weighted average convexities of the two bonds,
with weights beingthe present valueof the bond.
100
89
100
12
0.493
62
+
0.520
93
1.052
1.051
1.052
1.058
CB =
100
100
0.493 62 +
0.520 93
1.051
1.058
12
100
89
100
0.493 62 +
0.520 93
1.052
1.051
1.052
1.058
=
= 29. 024 9
100
1.054
CB > CA
If the interest rate moves
up by
P1A =
100
(1 0.037 2) = 79. 2098
1.054
P1B =
100
(1 0.036 6) = 79. 259 2
1.054
263
1. Two portfolios can have the same present value, the same duration, but
dierent convexities.
2. Under the assumption of the parallel shift of a flat yield curve, we can
always make free money by buying the highconvexity portfolio and sell
the lowconvexity portfolio.
3. The parallel shift of a flat yield curve assumption leads to arbitrage.
Information about three zerocoupon bonds:
Maturity (Yrs) Face
2
100
4
100
7
100
The current interest rate is 7% for all three bonds. Assume a parallel shift
of a flat yield curve. Design an arbitrage strategy.
We need to form 2 portfolios. These two portfolios have the same PV, the
same duration, but dierent convexity. We can make free money by buying the
high convexity portfolio and selling the low convexity portfolio.
The low convexity portfolio is the 4year bond (Portfolio A).
The high convexity portfolio consists of x unit of 2year bond and y unit
of 7year bond (Portfolio B). This is called a barbell. A barbell bond portfolio
combines short maturities (low duration) with long maturities (high duration)
for a blended, moderate maturity (moderate duration)
Portfolio
Maturity (Yrs)
PV
100
= 76. 29
1.074
100
= 87. 34
1.072
100
= 62. 27
1.077
100
100
100
x+
y=
1.072 1.077
1.074
100
100
100
x 2+
y 7=4
2
7
1.07
1.07
1.074
Duration
4
2
7
Convexity
45
= 17. 468 8
1.072
23
= 5. 240 6
1.072
78
= 48. 912 6
1.072
x = 0.524 1, y = 0.490 0
Units
1
x
y
264
CB
0.524 1 +
0.490 0
2
2
2
7
1.07
1.07
1.07
1.07
=
= 22. 708 9
100
1.074
For example, if the new interest rate is 7.25% for all bonds with dierent
maturities, then
100
100
P1B =
0.524 1 + 0.490 0
= 75. 584 1
1.07252
1.07257
100
= 75. 580 9
1.07254
B
A
P1 > P1
Our profit is 75. 584 1 75. 580 9 = 0.003 2
P1A =
If the new interest rate is 6.5% for all bonds with dierent maturities, then
100
100
P1B =
0.524 1 + 0.490 0
= 77. 739 6
2
1.065
1.0657
100
= 77. 732 3
1.0654
P1B > P1A
Our profit is 77. 739 6 77. 732 3 = 0.007 3
P1A =
24.1.2
We all know what an interest rate is, yet a derivative on interest rate is surprisingly dicult. To get a sense of the diculty, suppose we want to calculate the
price of a European call on a 1year zerocoupon bond that pays $100 one year
from now. Here are the inputs:
The call expires in one year
The strike price is $100
The continuous risksfree rate is r = 6% per year
The current price of the bond is 100e0.06 = 94. 18.
The volatility of the bond return is = 30%
Using the BlackScholes formula, we find:
d2 = 0.15
N (d1 ) = 0.5596
N (d2 ) = 0.4404
d1 = 0.15
C = 94. 18 0.5596 94. 18 0.4404 = 11. 23
265
Everything looks fine. However, after further thinking, you realize that the
call price C should be zero. At T = 1, the bond pays $100. Hence the 100strike
call value is zero. Why should anyone buy a 100strike call on an asset worth
$100 at call expiration?
What went wrong? It turns out that we cant use the BlackScholes formula
to calculate the price of an interest rate derivative:
The BlackScholes option formula assume that the term structure of the
interest rate is flat and deterministic (see DM page 379). However, If the
interest rate is known and constant, there wont be any need for interest
rate derivatives. This is similar to the idea that if the stock price is known
and constant, there wont be any need for call or put option.
The standard deviation of the return is a constant. However, the standard deviation of the return is not constant. Unlike a stock, a bond has
a finite maturity. At the maturity date, the bond value is its face amount.
Hence the standard deviation of a bonds return decreases as the bond
approaches its maturity.
The key point. Its much harder to calculate the price of an option interest
rate because interest rate is not a tradeable underlying asset. For call and put
on stocks, we can buy or short sell stocks to set up the hedge portfolio. However,
we cant go out and buy a 5% interest rate to hedge an option on interest rate.
Now lets go to the textbook.
24.1.3
A bond is a derivative on interest rate. We normally dont think this way, but
a bond derives its value from interest rate. If the market interest rate goes up,
the bond value goes down; if the market interest rate goes down, the bond value
goes up.
So our starting point is to set up a stochastic random variable called the
short interest rate. The textbook keeps using the phrase "short interest
rate" or "short rate" without giving a clear definition. Here is the definition:
Definition 24.1.1. A short interest rate or short rate is just the instantaneous
interest rate r (t) over a short (hence the name "short rate") interval [t, t + dt].
First, we assume that the short interest rate r (t) follows the Itos process:
dr = (r) dt + (r) dZ
(24.1)
Next, the text book explains that we cant assume r (t) follows a flat yield
curve ("impossible bond pricing model").
A flat yield curve means that the interest rate is independent of time. To
have a flat yield curve, the following two conditions are met:
266
The initial interest rate r is a constant regarding time, that is, r (t) = t
If the interest rate changes, the change is also independent of time (i.e.
parallel shift of the yield curve)
If the continuously compounded interest rate r is constant, then the present
value at time t of $1 to be received at T is:
P (t, T ) = er(T t)
(24.2)
267
r(T1 t)
= r
= er(tT1 ) (T1 t) = P (t, T1 ) (T1 t)
2e
2
2
(dr) = (dt + dZ) = 2 (dZ) = 2 dt
dP (t, T1 ) = rP (t, T1 ) dt P (t, T1 ) (T1 t) dr + 12 P (t, T1 ) (T1 t)2 2 dt
Similarly,
2
dP (t, T2 ) = rP (t, T2 ) dt P (t, T2 ) (T2 t) dr + 12 P (t, T2 ) (T2 t) 2 dt
However, dW = rW dt. To see why, notice that dW is the interest earned on
W during the interval [t, t + dt]. During [t, t + dt], the continuous interest rate
r can be treated as a discrete interest rate. The interest earned is just
the initial capital interest rate length of the interval = rW dt
Now wehhave DM Equation 24.6:
i
2
dI = rP (t, T1 ) dt P (t, T1 ) (T1 t) dr + 12 P (t, T1 ) (T1 t) 2 dt
h
i
+ rP (t, T2 ) dt P (t, T2 ) (T2 t) dr + 12 P (t, T2 ) (T2 t)2 2 dt
+rW dt
Well want to have dI = 0. If dI = 0, then the value of our portfolio wont
change during the same during [t, t + dt]. To make dI = 0, we first choose
such that the dr term is zero.
[P (t, T1 ) (T1 t)] + [P (t, T2 ) (T2 t)] = 0
2 t)P (t,T2 )
= (T
(T1 t)P (t,T1 )
units of bond
2 t)P (t,T2 )
By the way, = (T
(T1 t)P (t,T1 ) is similar to DM Equation 7.13 (Derivatives
Markets page 227):
N =
D1 B1 (y1 ) / (1 + y1 )
D2 B2 (y2 ) / (1 + y2 )
(DM 7.13)
268
24.1.4
Since its bad to assume that r is a flat yield curve, we switch our gears and
assume that r isnt a flat yield curve. Now we just assume that Equation 24.1
holds.
dP (r, t, T ) =
P
P
1 2P
(dr)2
dr +
dt +
r
t
2 t2
P
P
1 2P 2
P
(r)
+
dP (r, t, T ) = (r)
+
dt +
(r) dZ
r
2 t2
t
r
(24.3)
Define
(r, t, T ) =
P
P
1 2P 2
1
(r)
+
(r)
+
P (r, t, T )
r
2 t2
t
(24.4)
P
1
(r)
P (r, t, T ) r
(24.5)
q (r, t, T ) =
(24.6)
269
dP (r, t, T )
is the bonds return. Equation 24.6 says that the bonds return
P (r, t, T )
is the sum of a drift term (r, t, T ) dt and a random component q (r, t, T ) dZ.
P
is
Please also note that generally q (r, t, T ) is negative. This is because
r
negative. If r goes up, the bond price goes down; if r goes down, P goes up.
Next, we are going to derive DM Equation 24.16:
2 (r, t, T2 ) r
1 (r, t, T1 ) r
=
q1 (r, t, T1 )
q2 (r, t, T2 )
(24.7)
[1 (r, t, T1 ) r] dt + [2 (r, t, T2 ) r] dt = 0
q1 (r, t, T1 )
2 (r, t, T2 ) r
1 (r, t, T1 ) r
=
q1 (r, t, T1 )
q2 (r, t, T2 )
270
(24.8)
Apply Equation
24.4 and 24.6 to 24.8:
P
P
1 2P 2
1
(r)
+
(r)
+
r
P (r, t, T )
r
2 t2
t
= (r, t)
P
1
(r)
P (r, t, T ) r
P
1 2P 2
P
P
(r) +
+
rP = (r, t)
(r)
(r)
r
2 t2
t
r
2P
P
P
1 2
(r) 2 + [ (r) (r) (r, t)]
+
rP = 0
(24.9)
2
r
r
t
Any interestdependent securities (not just zero coupon bonds) must satisfy
Equation 24.9.
To solve the bond price P (r, t, T ), we need to use Equation 24.9 together
with the following boundary condition:
P (r, T, T ) = 1
(24.10)
P (r, t, T ) = Et
" Z
exp
#!
r (s) ds
(24.11)
where E means that the expectation is based on the risk neutral probability
and
Z
T
R (t, T ) =
r (s) ds
(24.12)
24.1.5
271
24.2
24.2.1
A simple model is to assume that the short rate r (t) follows arithmetic Brownian
motion:
dr (t) = dt + dZ
(24.14)
0
1 2
AB 2 eBr [ ] ABeBr A0 eBr = rA eBr B eBr
2
(24.15)
Equation 24.15 should hold for any r. The only way to make it work for any
r is:
1 2
2
0
2 AB [ ] AB A = 0
A 1B
=0
272
B =1
B =T t
From 12 2 AB 2 [ ] AB A0 = 0. we get:
A0
2
= 12 2 B 2 [ ] B = 12 2 (T t) [ ] (T t)
A
2
d ln A (T t) = 12 2 (T t) ( ) (T t)
1
3
2
ln A (T t) = 16 2 (T t) ( ) (T t) + C where C is a con2
stant.
1
3
2
1 2
A = A (0) exp 6 (T t) ( ) (T t)
2
1
3
2
1 2
= exp 6 (T t) ( ) (T t)
2
Finally, we have:
1
3
2
1 2
P (r, t, T ) = exp 6 (T t) ( ) (T t) e(T t)r(t)
2
You dont need to memorize the above formula. Just understand how to
derive the formula in case SOA or CAS gives you a tough problem.
24.2.2
RendlemanBartter model
(24.16)
Advantage:
r (t) cant be negative
V ar [r (t)] = r2 (t) 2 t. The variance increases if r (t) increases. This is
desirable because the volatility of the interest is high if the interest rate
is high.
Disadvantage:
not meanreverting
24.2.3
Vasicek model
According to Wikipedia, Vasiceks model was the first one to capture mean reversion, an essential characteristic of the interest rate that sets it apart from
other financial prices. Stock prices can rise indefinitely. However, interest rates
cannot. Excessively high interest rate would hamper economic activity, prompting a decrease in interest rates. Similarly, interest rates can not decrease indefinitely. If interest rates are too low, few people are willing to lend their money.
This tends to push up the interest rate. As a result, interest rates move in a
limited range, showing a tendency to revert to a long run value.
273
(24.17)
= AB 2 eBr
= ABeBr
= rAB eBr A0 eBr
2
r
r
t0
12 2 AB 2 eBr [ (b r) ] ABeBr +rAB eBr A0 eBr rAeBr =
0
0
12 2 AB 2 [ (b r) ] AB + rAB A0 rA = 0
0
12 2 AB 2 (b ) AB + AB A0 = A AB aAB r
For the above equation to hold for any r, we need to have:
0
A AB AB = 0
1 2
2
0
2 AB (b ) AB + AB A = 0
0
1 B B = 0
A AB AB = 0
The boundary condition is B (0) = 0
1 e(T t)
This gives us: B =
274
P (0.05, 1, 3) = 0.9321
P (r , 2, 4) = 0.8960
Calculate r .
The price of any bond must satisfy Equation 24.9. We guess that the solution
to Equation 24.9 is P (r, t, T ) = A (T t) eB(T t)r(t)
P (0.04, 0, 2) = 0.9445
A (2) eB(2)0.04 = 0.9445
P (0.05, 1, 3) = 0.9321
A (2) eB(2)0.05 = 0.9321
B(2)0.04
0.9445
0.9445
A (2) e
=
eB(2)0.01 =
B(2)0.05
0.9321
0.9321
A (2) e
1
100
0.9445 0.01
0.9445
eB(2) =
= 3. 749 26
=
0.9321
0.9321
0.04
r
= 0.8960
0.995 77 3. 749 26
0.8960
r
3. 749 26
=
= 0.899 81
0.995 77
r ln 3. 749 26 = ln 0.899 81
ln 0.899 81
r =
= 0.07988 4 = 0.08
ln 3.
749 26
Please note r can be solved without using any specifics of the Vasicek model.
As a matter of fact, this problem can be rewritten as:
Let P (r, t, T ) denote the price at time t of $1 to be paid with certainty at
time T , t T , if the short rate at time t is equal to r. You are given:
P (0.04, 0, 2) = 0.9445
P (0.05, 1, 3) = 0.9321
P (r , 2, 4) = 0.8960
Calculate r .
24.2.4
CIR model
The model assumes that the short interest rate r (t) follows the stochastic differential equation:
p
dr (t) = (b r) dt + r (t)dZ
(24.18)
In CIR model, V ar [r (t)] = 2 r (t) t.
The drift factor (b r) in the CIR model is the same as the drift factor
in the Vasicek model. It ensures mean reversion of the interest rate towards
the long run value b, with speed of adjustment governed by the strictly positive
parameter a.
p
The standard deviation factor, r (t), corrects the main drawback of Vasiceks model, ensuring that the interest rate cannot become negative. Thus,
at low values of the interest rate, the standard deviation becomes close to zero,
275
canceling the eect of the random shock on the interest rate. Consequently,
when the interest rate gets close to zero, its evolution becomes dominated by
the drift factor, which pushes the rate upwards.
Advantage:
Meanreverting
Not allow a negative interest rate
Variance of V ar [r (t)] = 2 r (t) t. The higher the r (t), the higher the
V ar [r (t)]. This is a desirable feature.
To solve the bond price under CIR model, again we guess the solution is
P (r, t, T ) = A (T t) eB(T t)r(t) . The solution is listed in DM page 788. You
dont need to memorize the solution. Just memorize P (r, t, T ) = A (T t) eB(T t)r(t) .
24.3
24.3.1
Black formula
Notations:
Pt (T, T + s). This is the price agreed upon at time t, which will be paid at
T in order to receive $1 at T + s. Simply put, Pt (T, T + s) is the present
value of $1 discounted from T + s to T using the interest rate available
at t. For example, at time zero we know that the annual interest rate
from t = 1 to t = 3 is 10% compounded continuously. Then P0 (1, 1 + 2)
is just PV of $1 discounted from t = 3 to t = 1 using 10% continuously
compounded interest rate. So P0 (1, 1 + 2) = e0.1(2) = 0.818 73
PT (T, T + s). This is PV $1 discounted from T + s to T . Generally, we
simplify the symbol PT (T, T + s) as P (T, T + s)
Consider a call option with strike price K, expiring at time T , on a zerocoupon bond paying $1 at time T + s. If you buy this call option, then at T you
have the right to buy a bond maturing at time T + s for the guaranteed price
K. Since time T cost of a bond maturing in T + s is PV of $1 discounted from
T + s to T , buying a bond maturing at time T + s for the guaranteed price K
really means "give up K and receive PV of $1 discounted from T + s to T ." The
PV of $1 discounted from T + s to T is PT (T, T + s). So the call payo is
Call P ayof f = max [0, PT (T, T + s) K]
(DM 24.30)
P (t, T + s)
P (t, T )
(DM 24.31)
276
d1 =
P (S)
F0,T
+0.5 2 T
F P (K)
0,T
d2 = d1 T
d2 = d1 T
277
C =Time zero cost of what you get at T N (d1 ) Time zero cost of what
you giveat T N (d2 )
Time zero cost of what you get at T
2
d1 = ln Time
+
0.5
T
/ T
zero cost of what you give at T
d2 = d1 T
ln
P (0,T +s)
2
P (0,T ) +0.5 T
d2 = d1 T
ln
2
F
T
K +0.5
278
ln
2
0.8817
0.92590.9434 +0.50.05 1
= 0.212 0
0.05 1
N (d2 ) = 0.564 3
N (d1 ) = 0.583 9
C = 0.8817 0.583 9 0.9259 0.9434 0.564 3 = 0.021 9
We can also calculate the put price.
P =Time zero cost of what you give at T N (d2 ) Time zero cost of what
you get at T N (d1 )
P = 0.9259 0.9434 (1 0.564 3) 0.8817 (1 0.583 9) = 0.013 7
If you want to use the formula C = P (0, T ) [F N (d1 ) KN (d2 )], this is
how:
P (0, T ) = P (0, 1) is PV of $1 discounted from T = 1 to time zero. So
P (0, 1) = 0.9434
F = F0,T [P (T, T + s)] = F0,1 [P (1, 2)] is the forward price of 1year bond.
This is the price agreed up at time zero, paid at T = 1 in order to receive
$1 at T + s = 2. Using Equation DM 24/31, we have F0,1 [P (1, 2)] =
P (0,2)
0.8817
P (0,1) = 0.9434
d1 =
ln
F
K
+0.5 2 T
ln
0.8817
0.9434
0.9259
+0.50.052 1
0.05 1
= 0.212 0
= 0.9434 0.8817
0.9434 0.583 9 0.9259 0.564 3 = 0.0220
0.8817
0.9434
(1 0.583 9)
24.3.2
279
Notation
Rt (T, T + s). The (not annualized) interest rate preagreed upon at time
t where t T that applies to the future time interval [T, T + s].
RT (T, T + s). The (not annualized) interest rate agreed upon at time T
that applies to the time interval [T, T + s].
Caplet. A caplet gives the buyer the right to buy the timeT market
interest rate RT (T, T + s) by paying a fixed strike interest rate KR . If
KR RT (T, T + s), the caplet expires worthless. The payo of the caplet
at T + s is max [0, RT (T, T + s) KR ]. The payo of the caplet at T is
max [0, RT (T, T + s) KR ]
1 + RT (T, T + s)
To calculate the price of the caplet, we first modify
the payo:
1
1
1
PT (T, T + s)
(1 + KR ) max 0,
= (1 + KR ) max 0,
1 + KR
1 + RT(T, T + s)
1 + KR
1
max 0,
PT (T, T + s) is the payo of a put on a bond. This put
1 + KR
gives the buyer the right, at T , to sell a bond that matures at T + s for a
1
guaranteed price 1+K
. Let P represent the price of this put. Hence the price
R
of the caplet is (1 + KR ) P .
24.4
Binomial interest rate tree is really simple. Unfortunately, the author of the
textbook uses too many math symbols and formulas, making this section hard
to read. What I will do here is to walk you through a few examples. If you
understand these examples, you are fine.
Example 24.4.1. (DM Example 24.3)
The following is the 3period interest rate tree (DM Figure 24.3)
t=0 t=1 t=2
0.18
0.14
0.10
0.10
0.10
0.06
0.02
280
Make sure you understand the above table. The 10% interest rate at t = 0
applies to the interval [t = 0, t = 1]. The 14% and 6% interest rates apply to
the interval [t = 1, t = 2]. The interest rates 0.18, 0.10, and 0.02 at t = 2 apply
to the interval [t = 2, t = 3].
The price of the 1year bond is just the PV of $1 discounted from t = 1 to
t = 0. Hence P (0, 1) = e0.1
The price of the 2year bond is just the PV of $1 discounted from t = 2 to
t = 0. Whats tricky is that we have two interest rates during [t = 1, t = 2].
If the path is u, then $1 at t = 2 travels back to t = 0 through 2 interest
rates: 0.14, and 0.1. The PV of $1 discounted from t = 2 to t = 1 is
e0.14 . The PV of e0.14 discounted from t = 1 to t = 0 is e0.14 e0.11 =
e(0.14+0.1) . So the 2year bond is worth e(0.14+0.1) at t = 0
If the path is d, then $1 at t = 2 travels back to t = 0 through 2 interest
rates: 0.06 and 0.1. The 2year bond is worth e(0.06+0.1) at t = 0
Since the riskneutral probability of up or down is 50%, then the 2year bond
is worth the following at t = 0:
P (0, 2) = 0.5e(0.14+0.1) + 0.5e(0.06+0.1) = 0.819 4
The term 0.5e(0.14+0.1) + 0.5e(0.06+0.1)
can be irewritten as:
h S
(0.14+0.1)
(0.06+0.1)
2i=0 ri h
0.5e
+ 0.5e
=E e
with h = 1
The price of the 3year bond is just the PV of $1 discounted from t = 3 to
t = 0.
If the path is uu, then $1 at t = 3 travels back to t = 0 through 3 interest
rates: 0.18, 0.14, and 0.1. The PV of $1 discounted from t = 3 to t = 0 is
e(0.18+0.14+0.1) . So the 3year bond is worth e(0.18+0.14+0.1) at t = 0
If the path is ud, then $1 at t = 3 travels back to t = 0 through 3 interest
rates: 0.1, 0.14, and 0.1. The 3year bond is worth e(0.1+0.14+0.1) at t = 0
If the path is du, then $1 at t = 3 travels back to t = 0 through 3 interest
rates: 0.1, 0.06, and 0.1. The 3year bond is worth e(0.1+0.06+0.1) at t = 0
If the path is dd, then $1 at t = 3 travels back to t = 0 through 3 interest
rates: 0.02, 0.06, and 0.1. The 3year bond is worth e(0.02+0.06+0.1) at
t=0
Since the riskneutral probability for each is 0.52 = 0.25, the price of a 3year
bond is the following
at t = 0:
281
(0.18+0.14+0.1)
h Sn
i
P (0, n) = E e i=0 ri h
(DM 24.31)
282
As a borrower, you are worried that the market interest may go up. For
example, if the Year 1 interest rate is 20%, then your interest payment at the
end of Year 1 is 100 0.2 = 20.
How can you reduce your risk? One thing you can do is to buy an interest
cap. Suppose you buy an interest rate cap of 7.5%. This is what happens:
If the market interest rate is at or below 7.5%, then the cap doesnt kick
in. So you get nothing from the cap
if the market interest rate is above 7.5%, then the party who sold you
the cap will pay you the excess of the market interest rate over the cap
rate. For example, if the market interest rate in Year 1 is 10%, then the
seller of the cap will pay you 100 (0.1 0.075) = 2. 5 at the end of Year
1. You still own the bank 100 0.1 = 10 at the end of Year 1, but you
pay 100 0.075 = 7. 5 out of your own pocket. The cap seller pays you
100 (0.1 0.075) = 2. 5. So together you get 7. 5 + 2.5 = 10. You mail a
$10 check to the bank.
In summary, if you buy an interest capped of 7.5%, then the interest on
your loan is capped at 7.5% regardless of the market interest rate (because any
excess of the market rate over the cap 7.5% is paid by the cap seller).
Now you know what an interest cap is, lets solve this problem.
The first year market rate 6% is below the cap rate. At the end of Year 1,
you get nothing from the cap
If the 2nd year market rate is 7.704%, then at the end of year 2 the cap
seller pays you 100 (0.07704 0.075) = 0.204 . Notice that 0.204 occurs at
t = 2. To help keep track of payments, well discount this payment to t = 1.
0.204
The discounted value is
at t = 1.
1 + 0.07704
If the 2nd year market rate is 4.673%, then the payment at the end of year
2 is zero.
Lets calculate the cap payo in Year 3. Of the 4 interest rates during
[t = 2, t = 3] , only when the market interest rate is 9.892% do we get a payo
of 100 (0.09892 0.075) = 2. 392. This payment occurs at t = 3. The PV of
2. 392
.
this payment at t = 2 is
1 + 0.09892
Now we can draw a payo table:
t=1
283
t=2
9.892% Payo:
7.704% Payo:
6.000%
6%
6.000%
4.673%
3.639%
100 (0.07704 0.075)
= 0.189 4 1 at t = 1
1 + 0.07704
100 (0.07704 0.075)
t = 0 is
. The risk neutral probability
(1 + 0.09892) (1 + 0.06)
node is 0.5.
100 (0.09892 0.075)
= 2. 176 68 at t = 2
The PV of payo
1 + 0.09892
100 (0.09892 0.075)
t = 0 is
. The risk neutral
(1 + 0.09892) (1 + 0.07704) (1 + 0.06)
reaching uu node is 0.52 = 0.25
The PV of payo
discounted to
of reaching u
discounted to
probability of
Hence risk neutral based expected present value of the cap payo is:
100 (0.07704 0.075)
100 (0.09892 0.075)
0.5+
0.52 =
(1 + 0.07704) (1 + 0.06)
(1 + 0.09892) (1 + 0.07704) (1 + 0.06)
0.565 99 = 0.57
So the price of the cap is $0.57
24.5
BlackDermanToy model
284
YTM (yield to maturity) is the discrete annual eective interest rate earned
by a bond. The formula is
P (0, T ) = (1 + Y T M )T
For example, a 2year bond in the table is worth 0.8116. This means that
PV of $1 discounted from t = 2 to t = 0 is 0.8116. To find YTM, we solve the
following equation:
2
0.8116 = (1 + i)
i = 0.11
So the YTM for this 2year bond is 11%
The volatility in Yr 1 on (n 1) bond is the standard deviation of the natural
log of the YTM on an (n 1)year bond issued at t = 1 and maturity at time
n. For example, 10% volatility in the table means that standard deviation of
the natural log of the YTM on a 1year bond issued at t = 1 maturing in t = 2
is 10%. This concept will be clear to you later.
Next, next use the BDT model to build an interest rate tree. Well first find
the interest rate ru and rd , the two possible interest rates for Year 2 (i.e. for
the time interval [t = 1, t = 2]).
t=0
t=h=1
ru
r0 = 10%
rd
Once again, rt is the interest for the interval [t, t + 1]. For example, r0 = 10%
is the interest rate for Year 1. ru and rd are the two interest rates for Year 2
(i.e. from t = 1 to t = 2).
We assume the risk neutral probability of up and down is 0.5. We want to
find ru and rd to satisfy the condition that standard deviation of the natural
log of the YTM on a 1year bond issued at t = 1 maturing in t = 2 is 10%.
The price of a 1year bond issued at t = 1 maturing in t = 2 is
1
1
P (1, 2) =
or
1 + ru
1 + rd
The YTM is:
1
1
Y T Mu = ru
(1 + Y T Mu ) =
1 + ru
1
1
or (1 + Y T Md ) =
Y T Md = rd
1 + rd
We know that the standard deviation of ln Y T Mu = ln ru and ln Y T Md =
ln rd is 10%.
The mean of the log of the YTM on a 1year bond issued at t = 1 maturing
in t = 2 rate is:
E (ln r1 ) = 0.5 ln ru + 0.5 ln rd = 0.5 (ln ru + ln rd )
285
2
2
ru
2
= (0.5 ln ru 0.5 ln rd ) = 0.5 ln
rd
The standard deviation of ln Y T Mu = ln ru and ln Y T Md = ln rd is 1 =
10%.
ru
ru
= 1
ln
= 2 1
ru = rd e21 = rd e0.2
0.5 ln
rd
rd
Now we see that matching volatility requires ru = rd e2 . This relationship
holds for every node.
Next, we want to reproduce the 2year bond price of 0.8116.
If the Year 2 interest rate is ru , then the PV of $1 discounted from t = 2 to
1
t = 0 is
(1 + ru ) (1 + r0 )
If the Year 2 interest rate is rd , then the PV of $1 discounted from t = 2 to
1
t = 0 is
(1 + rd ) (1 + r0 )
The riskneutral probability of up and down is 0.5. The expected 2year
bond price is:
1
1
0.5
+ 0.5
(1 + ru ) (1 + r0 )
(1 + rd ) (1 + r0 )
To match the observed price P (0, 2) = 0.8116, we have:
1
1
0.5
+ 0.5
= P (0, 2)
(1 + ru ) (1 + r0 )
(1 + rd ) (1 + r0 )
To
sum up, we have two equations:
ru = rd e21
1
1
+ 0.5
= P (0, 2)
0.5
(1 + ru ) (1 + r0 )
(1 + rd ) (1 + r0 )
Or
ru = rd e0.2
0.5
1
1
+ 0.5
= 0.8116
(1 + ru ) (1 + 0.1)
(1 + rd ) (1 + 0.1)
1
1
+
= 2 0.8116 (1 + 0.1)
1 + rd 1 + rd e0.2
rd = 0.108 265 = 10.83%
286
1
0.5
1
+
=
1.1322 1 + rdd e42
1 + rdd e22
The YTM can solved as follows:
P (1, 3, ru ) = (1 + i)2
i = P (1, 3, ru )0.5 1
If the 2nd year interest rate is rd = 10.83%, then the 3rd year rate is either
rdu or rdd with equal riskneutral probability of 0.5. Then the expected PV of
$1 discounted from t = 3 to t = 1 is
1
1
P (1, 3, rd ) = 0.5
+ 0.5
(1 + rd ) (1 + r
(1 + rd ) (1 + rdd )
du )
1
0.5
1
+
=
1.1083 1 + rdd e22
1 + rdd
The YTM can solved as follows:
2
P (1, 3, rd ) = (1 + i)
i = P (1, 3, rd )
0.5
0.5
P (1, 3, ru )
1
0.15 = 0.5 ln
0.5
P (1, 3, rd )
1
0.5
1
0.5
1
+
1
1.1322 1 + rdd e42
1 + rdd e22
= 0.5 ln
0.5
0.5
1
1
+
1
2
2
1.1083 1 + rdd e
1 + rdd
By the way, please note that 2 is not 10%.
287
1
if the path is 0 u uu
(1 + r0 ) (1 + ru ) (1 + ruu )
1
if the path is 0 u ud
(1 + r0 ) (1 + ru ) (1 + rud )
1
if the path is 0 d du
(1 + r0 ) (1 + rd ) (1 + rdu )
1
if the path is 0 d dd
(1 + r0 ) (1 + rd ) (1 + rdd )
1
1
0.5
+
1
1.1322 1 + rdd e42
1 + rdd e22
0.15 = 0.5 ln
0.5
1
0.5
1
+
1
1.1083 1 + rdd e22
1 + rdd
0.25
0.25
+
4
2
(1.1) (1.1322) (1 + rdd e ) (1.1) (1.1322) (1 + rdd e22 )
0.25
0.25
+
+
= 0.7118
2
2
(1.1) (1.1083) (1 + rdd e ) (1.1) (1.1083) (1 + rdd )
These equations are hard to solve manually. Special software is needed to
solve them.
2 = 0.195 0
You can verify that the solutions are: rdd = 0.0925
0.5
1
1
0.5
+
1
1.1322 1 + 0.0925e40.1950 1 + 0.0925e20.1950
= 0.149 97 =
0.5 ln
0.5
0.5
1
1
+
1
1.1083 1 + 0.0925e20.1950 1 + 0.0925
0.15
288
0.25
0.25
+
(1.1) (1.1322) (1 + 0.0925e40.1950 ) (1.1) (1.1322) (1 + 0.0925e20.1950 )
0.25
0.25
+
+
(1.1) (1.1083) (1 + 0.0925e20.1950 ) (1.1) (1.1083) (1 + 0.0925)
= 0.711 755 = 0.7118
www.actuary88.com
c
Yufeng
Guo
ii
Contents
Introduction
vii
23
91
12 BlackScholes formula
107
125
14 Exotic options: I
149
18 Lognormal distribution
161
177
187
193
22 Exotic options: II
203
23 Volatility
205
209
iii
CONTENTS
www.actuary88.com
CONTENTS
c
Yufeng
Guo
iv
Preface
This is Guos solution to Derivatives Markets (2nd edition ISBN 032128030X)
for SOA MFE or CAS Exam 3 FE. Unlike the ocial solution manual published
by AddisonWesley, this solution manual provides solutions to both the evennumbered and oddnumbered problems for the chapters that are on the SOA
Exam MFE and CAS Exam 3 FE syllabus. Problems that are out of the scope
of the SOA Exam MFE and CAS Exam 3 FE syllabus are excluded.
Please report any errors to yufeng_guo@msn.com.
This book is the exclusive property of Yufeng Guo. Redistribution of this
book in any form is prohibited.
PREFACE
www.actuary88.com
PREFACE
c
Yufeng
Guo
vi
Introduction
Recommendations on using this solution manual:
1. Obviously, youll need to buy Derivatives Markets (2nd edition) to see the
problems.
2. Make sure you download the textbook errata from http://www.kellogg.
northwestern.edu/faculty/mcdonald/htm/typos2e_01.html
vii
CHAPTER 8. INTRODUCTION
www.actuary88.com
c
Yufeng
Guo
viii
Chapter 9
T = 6/12 = 0.5
K = 35
r = 0.04
= 0.06
C + P V (K) = P + S0 eT
2.27 + 35e0.04(0.5) = P + 32e0.06(0.5)
P = 5. 522 7
Problem 9.2.
S0 = 32
T = 6/12 = 0.5
K = 30
C = 4.29
P = 2.64
r = 0.04
C + P V (K) = P + S0 P V (Div)
4.29 + 30e0.04(0.5) = 2.64 + 32 P V (Div)
P V (Div) = 0.944
Problem 9.3.
S0 = 800
r = 0.05
=0
T =1
K = 815
C = 75
P = 45
a. Buy stock+ sell call+buy put=buy P V (K)
C + P V (K) = P + S0
P V (K = 815) = S0 + C + {z}
P = 800 + (75) + 45 = 770
{z}
{z}
buy stock
sell call
buy put
So the position is equivalent to depositing 770 in a savings account (or buying a bond with present value equal to 770) and receiving 815 one year later.
770eR = 815
R = 0.056 8
1
sell call
buy put
c
Yufeng
Guo
1 1
C$ (x0 , K, T ) = x0 KPf
, ,T
x0 K
Just use my approach to solve this type of problems.
Convert information to symbols:
1
The exchange rate is 95 yen per euro. Y 95 =1 or Y 1 =
95
Yendenominated put on 1 euro with strike price Y100 has a premium Y8.763
(1 Y 100)0 =Y8.763
Whats the strike price of a eurodenominated call on 1 yen? K 1Y
Calculate the price of a eurodenominated call on 1 yen with strike price K
(K 1Y )0 = ?
1
Y1
100
1
The strike price of the corresponding eurodenominated yen call is K =
=0.01
100
1
1
1
Y1 =
(1 Y 100)0 =
(Y 8.763)
100
100
100
0
1 Y 100
1
, we have:
95
1
1
1
(Y 8.763) =
(8.763)
=9. 224 2 104
100
100
95
1
Y 1 =9. 224 2 104
100
0
Since Y 1 =
www.actuary88.com
c
Yufeng
Guo
1
100
b. There are two puts out there. One is the synthetically created put using
the formula:
P = C + KerT S0 eT
The other is the put in the market selling for the price for $0.0004.
To arbitrage, build a put a low cost and sell it at a high price. At t = 0, we:
Sell the expensive put for $0.0004
Build a cheap put for $0.00025. To build a put, we buy a call, deposit
KerT in a savings account, and sell eT unit of Yen.
t=0
0.0004
0.0006
0.009e0.05(1)
0.009e0.01(1)
$0.00015
T =1
ST < 0.009
ST 0.009
0
0.009
ST
0
T =1
ST 0.009
0
ST 0.009
0.009
ST
0
c
Yufeng
Guo
1
$1 =?
premium: Y
0.009
0
Well find the premium for Y 1 $0.009, the option of "give 1 yen and get
$0.009." Once we find this premium, well scale it and find the premium of "give
1
yen and get $1."
0.009
Well use the general putcall parity:
(AT BT )0 + P V (AT ) = (BT AT )0 + P V (BT )
($0.009 Y 1)0 + P V ($0.009) = (Y 1 $0.009)0 + P V (Y 1)
P V ($0.009) = $0.009e0.05(1)
Since we are discounting $0.009 at T = 1 to time zero, we use the dollar
interest rate 5%.
P V (Y 1) = $0.009e0.01(1)
If we discount Y1 from T = 1 to time zero, we get e0.01(1) yen, which is
equal to $0.009e0.01(1) .
So we have:
$0.0006+$0.009e0.05 = (Y 1 $0.009)0 + $0.009e0.01(1)
(Y
= $2. 506 16 104
1 $0.009)0
1
1
2. 506 16 104
=
Y 1 $1
(Y 1 $0.009)0 = $
= $2.
0.009
0.009
0.009
0
2
2. 784 62 10
= Y 3. 094
784 62 102 = Y
0.009
www.actuary88.com
c
Yufeng
Guo
$Y
1
0.009
1
($0.009 Y 1)0
0.009
1
=
$0.0006 = $ 0.0 666 7
0.009
1
= Y 7. 407 8
= Y 0.0 666 7
0.009
So the yen denominated atthemoney put for $1 is worth $ 0.0 666 7 or Y
7. 407 8.
I recommend that you use my solution approach, which is less prone to errors
than using complex notations and formulas in the textbook.
Problem 9.8.
The textbook Equations 9.13 and 9.14 are violated.
This is how to arbitrage on the calls. We have two otherwise identical
calls, one with $50 strike price and the other $55. The $50 strike call is more
valuable than the $55 strike call, but the former is selling less than the latter.
To arbitrage, buy low and sell high.
We use T to represent the common exercise date. This definition works
whether the two options are American or European. If the two options are
American, well find arbitrage opportunities if two American options are exercised simultaneously. If the two options are European, T is the common
expiration date.
The payo is:
Transaction
Buy 50 strike call
Sell 55 strike call
Total
t=0
9
10
1
T
ST < 50
0
0
0
T
50 ST < 55
ST 50
0
ST 50 0
T
ST 55
ST 50
(ST 55)
5
c
Yufeng
Guo
T
ST 55
ST 55
(ST 50)
5
c
Yufeng
Guo
K3 = 60
9.50
50 + (1 ) 60 = 55
= 0.5
0.5 (50) + 0.5 (60) = 55
Lets check:
C [0.5 (50) + 0.5 (60)] = C (55) = 14
0.5C (50) + 0.5C (60) = 0.5 (18) + 0.5 (9.50) = 13. 75
C [0.5 (50) + 0.5 (60)] > 0.5C (50) + 0.5C (60)
So arbitrage opportunities exist. To arbitrage, we buy low and sell high.
The cheap asset is the diversified portfolio consisting of units of K1 strike
option and (1 ) units of K3 strike option. In this problem, the diversified
portfolio consists of half a 50strike call and half a 60strike call.
The expensive asset is the 55strike call.
www.actuary88.com
c
Yufeng
Guo
T
50 ST < 55
T
55 ST < 60
T
ST 60
18
9.5
27. 5
0
0
0
ST 50
0
ST 50
ST 50
0
ST 50
ST 50
ST 60
2ST 110
2 (14) = 28
0.5
0
0
0
ST 50 0
2 (ST 55)
60 ST > 0
2 (ST 55)
0
Transaction
buy two portfolios
buy a 50strike call
buy a 60strike call
Portfolio total
t=0
27. 5 + 28 = 0.5
ST 50 2 (ST 55) = 60 ST
2ST 110 2 (ST 55) = 0
So we get $0.5 at t = 0, yet we have non negative cash flows at the expiration
date T . This is arbitrage.
The above strategy of buying units of K1 strike call, buying (1 ) units
of K3 strike call, and selling one unit of K2 strike call is called the butterfly
spread.
We are given the following 3 puts:
Strike
K1 = 50 K2 = 55
Put premium 7
10.75
K3 = 60
14.45
50 + (1 ) 60 = 55
= 0.5
0.5 (50) + 0.5 (60) = 55
Lets check:
P [0.5 (50) + 0.5 (60)] = P (55) = 10.75
0.5P (50) + 0.5P (60) = 0.5 (7) + 0.5 (14.45) = 10. 725
P [0.5 (50) + 0.5 (60)] > .5P (50) + 0.5P (60)
So arbitrage opportunities exist. To arbitrage, we buy low and sell high.
The cheap asset is the diversified portfolio consisting of units of K1 strike
put and (1 ) units of K3 strike put. In this problem, the diversified portfolio
consists of half a 50strike put and half a 60strike put.
The expensive asset is the 55strike put.
www.actuary88.com
c
Yufeng
Guo
Transaction
buy two portfolios
buy a 50strike put
buy a 60strike put
Portfolio total
t=0
T
ST < 50
T
50 ST < 55
T
55 ST < 60
T
ST 60
7
14.45
21. 45
50 ST
60 ST
110 2ST
0
60 ST
60 ST
0
60 ST
60 ST
0
0
0
2 (55 ST )
0
2 (55 ST )
ST 50 0
0
60 ST > 0
0
0
Problem 9.11.
This is similar to Problem 9.10.
We are given the following 3 calls:
Strike
K1 = 80 K2 = 100
Call premium 22
9
K3 = 105
5
80 + 105 (1 ) = 100
= 0.2
0.2 (80) + 0.8 (105) = 100
C [0.2 (80) + 0.8 (105)] = C (100) = 9
0.2C (80) + 0.8C (105) = 0.2 (22) + 0.8 (5) = 8. 4
C [0.2 (80) + 0.8 (105)] > 0.2C (80) + 0.8C (105)
So arbitrage opportunities exist. To arbitrage, we buy low and sell high.
The cheap asset is the diversified portfolio consisting of units of K1 strike
option and (1 ) units of K3 strike option. In this problem, the diversified
portfolio consists of 0.2 unit of 80strike call and 0.8 unit of 105strike call.
The expensive asset is the 100strike call.
www.actuary88.com
c
Yufeng
Guo
10
T
80 ST < 100
Transaction
buy ten portfolios
buy two 80strike calls
buy eight 105strike calls
Portfolio total
t=0
2 (22)
8 (5)
84
0
0
0
2 (ST 80)
0
2 (ST 80)
10 (9)
6
0
0
0
2 (ST 80) 0
Transaction
buy ten portfolios
buy two 80strike calls
buy eight 105strike calls
Portfolio total
t=0
T
100 ST < 105
T
ST 105
2 (22)
8 (5)
84
2 (ST 80)
0
2 (ST 80)
2 (ST 80)
8 (ST 105)
10ST 1000
10 (ST 100)
8 (105 ST ) > 0
10 (ST 100)
0
84 + 10 (9) = 84 + 90 = 6
2 (ST 80) + 8 (ST 105) = 10ST 1000
2 (ST 80) 10 (ST 100) = 840 8ST = 8 (105 ST )
10ST 1000 10 (ST 100) = 0
So we receive $6 at t = 0, yet we dont incur any negative cash flows at
expiration T . So we make at least $6 free money.
We are given the following 3 put:
Strike
K1 = 80 K2 = 100
Put premium 4
21
K3 = 105
24.8
80 + 105 (1 ) = 100
= 0.2
0.2 (80) + 0.8 (105) = 100
P [0.2 (80) + 0.8 (105)] = P (100) = 21
0.2P (80) + 0.8P (105) = 0.2 (4) + 0.8 (24.8) = 20. 64
www.actuary88.com
c
Yufeng
Guo
11
Transaction
buy ten portfolios
buy two 80strike puts
buy eight 105strike puts
Portfolio total
t=0
T
ST < 80
T
80 ST < 100
2 (4)
8 (24.8)
84
2 (80 ST )
8 (105 ST )
1000 10ST
0
8 (105 ST )
8 (105 ST )
10 (21)
3. 6
10 (100 ST )
0
10 (100 ST )
2 (ST 80) 0
Transaction
buy ten portfolios
buy two 80strike puts
buy eight 105strike puts
Portfolio total
t=0
T
100 ST < 105
T
ST 105
2 (4)
8 (24.8)
84
0
8 (105 ST )
8 (105 ST )
0
0
0
10 (21)
3. 6
0
8 (105 ST ) > 0
0
0
c
Yufeng
Guo
12
K1 = 90
10
K2 = 95
4
C (K1 ) C (K2 ) = 10 4 = 6
K2 K1 = 95 90 = 5
C (K1 ) C (K2 ) > K2 K1 P V (K2 K1 )
Arbitrage opportunities exist.
To arbitrage, we buy low and sell high. The cheap call is the 95strike call;
the expensive call is the 90strike call.
We use T to represent the common exercise date. This definition works
whether the two options are American or European. If the two options are
American, well find arbitrage opportunities if two American options are exercised simultaneously. If the two options are European, T is the common
expiration date.
The payo is:
T
T
T
Transaction
t = 0 ST < 90 90 ST < 95
ST 95
Buy 95 strike call 4
0
0
ST 95
Sell 90 strike call 10
0
(ST 90)
(ST 90)
Total
6
0
(ST 90) 5 5
We receive $6 at t = 0, yet we our max liability at T is 5. So well make
at least $1 free money.
b.
T =2
r = 0.1
Strike
K1 = 90
Call premium 10
K2 = 95
5.25
c
Yufeng
Guo
13
K3 = 105
6
1
90 + (1 ) 105 = 100
=
3
1
2
(90) + (105) = 100
3
3
2
1
(90) + (105) = C (100) = 10
C
3
3
2
1
2
1
C (90) + C (105) = (15) + (6) = 9
3
3
3
3
2
1
2
1
(90) + (105) > C (90) + C (105)
C
3
3
3
3
Hence arbitrage opportunities exist. To arbitrage, we buy low and sell high.
1
The cheap asset is the diversified portfolio consisting of unit of 90strike
3
2
call and unit of 105strike call.
3
The expensive asset is the 100strike call.
Since we cant buy a partial option, well buy 3 units of the portfolio (i.e.
buy one 90strike call and two 105strike calls). Simultaneously,we sell three
100strike calls.
The payo at expiration T :
T
ST < 90
T
90 ST < 100
T
100 ST < 105
T
ST 105
15
2 (6)
27
0
0
0
ST 90
0
ST 90
ST 90
0
ST 90
ST 90
2 (ST 105)
3ST 300
3 (10)
3
0
0
0
ST 90 0
3 (ST 100)
2 (105 ST ) > 0
3 (ST 100)
0
Transaction
buy 3 portfolios
buy one 90strike call
buy two 105strike calls
Portfolios total
t=0
c
Yufeng
Guo
14
c
Yufeng
Guo
15
option is at least as
Markets Page 298.
C1 = 11.50
C2 = 11.924
The longerlived call is cheaper than the shorterlived call, leading to arbitrage opportunities. To arbitrage, we buy low (Call #1) and sell high (Call
#2).
The payo at expiration T1 = 1.5 if ST2 < 100e0.05 = 105. 127
Transaction
Sell Call #2
buy Call #1
Total
t=0
11.924
11.50
0.424
T2
0
T1
ST1 < 100e0.05(1.5)
0
0
0
T1
ST1 100e0.05(1.5)
0
ST1 100e0.05(1.5)
ST1 100e0.05(1.5) 0
t=0
11.924
11.50
0.424
T2
100e0.05 ST2
c
Yufeng
Guo
T1
ST1 < 100e0.05(1.5)
100e0.05(1.5) ST1
0
100e0.05(1.5) ST1 < 0
T1
ST1 100e0.05(1.5)
100e0.05(1.5) ST1
ST1 100e0.05(1.5)
0
16
t=0
C a
KerL
Pb
S0b
P b + S0b (C a + KerL )
If ST < K
If ST K
0
K
ST K
ST
0
ST K
K
0
ST
0
c
Yufeng
Guo
17
t=0
Cb
KerB
P a
a
S
0b
C + KerB P b + S0b
If ST < K
If ST K
0
K
K ST
ST
0
K ST
K
0
ST
0
The
b payoris always
zero. To avoid arbitrage, we need to have
C + Ke B P b + S0b 0
Problem 9.17.
a. According to the putcall parity, the payo of the following position is
always zero:
1. Buy the call
2. Sell the put
3. Short the stock
4. Lend the present value of the strike price plus dividend
The existence of the bidask spread and the borrowinglending rate dierence
doesnt change the zero payo of the above position. The above position always
has a zero payo whether theres a bidask spread or a dierence between the
borrowing rate and the lending rate.
If there is no transaction cost such as a bidask spread, the initial gain of
the above position is zero. However, if there is a bidask spread, then to avoid
arbitrage, the initial gain of the above position should be zero or negative.
The
b initial
gain of the position is:
P + S0b [C a + P VrL (K) + P VrL (Div)]
Theres
no
b
arbitrage if
P + S0b [C a + P VrL (K) + P VrL (Div)] 0
In this problem, we are given
rL = 0.019
rB = 0.02
S0b = 84.85. We are told to ignore the transaction cost. In addition, we
are given that the current stock price is 84.85. So S0b = 84.85.
The dividend is 0.18 on November 8, 2004.
www.actuary88.com
c
Yufeng
Guo
18
a
K T
Ca
Pb
+ P VrL (Div)]
P + S0b [C
+ P VrL (K)
75 0.0986 10.3 0.2
0.2 + 84.85 10.3 + 75e0.0190.0986 + 0.18
= 0.29
0.0190.0986
80 0.0986 5.6
0.6
0.6 + 84.85 5.6 + 80e
+ 0.18 = 0.18
85 0.0986 2.1
2.1
2.1 + 84.85 2.1 + 85e0.0190.0986 + 0.18 = 0.17
90 0.0986 0.35 5.5
5.5 + 84.85 0.35 + 90e0.0190.0986 + 0.18 = 1. 15
0.0190.271 2
75 0.271 2 10.9 0.7
0.7 + 84.85 10.9
+ 75e0.0190.271 2 + 0.18 = 0.14
80 0.271 2 6.7
1.45 1.45 + 84.85 6.7 + 80e
+ 0.18
= 0.17
85 0.271 2 3.4
3.1
3.1 + 84.85 3.4 + 85e0.0190.271 2 + 0.18 = 0.19
90 0.271 2 1.35 6.1
6.1 + 84.85 1.35 + 90e0.0190.271 2 + 0.18 = 0.12
c
Yufeng
Guo
19
T
0.0986
0.0986
0.0986
0.0986
0.271 2
0.271 2
0.271 2
0.271 2
Cb
9.9
5.3
1.9
0.35
10.5
6.5
3.2
1.2
Pa
0.25
0.7
2.3
5.8
0.8
1.6
3.3
6.3
Problem 9.18.
Suppose there are 3 options otherwise identical but with dierent strike price
K1 < K2 < K3 where K2 = K1 + (1 ) K2 and 0 < < 1.
Then the price of the middle strike price K2 must not exceed the price of a
diversified portfolio consisting of units of K1 strike option and (1 ) units
of K2 strike option:
C [K1 + (1 ) K3 ] C (K1 ) + (1 ) C (K3 )
P [K1 + (1 ) K3 ] P (K1 ) + (1 ) P (K3 )
The above conditions are called the convexity of the option price with respect
to the strike price. They are equivalent to the textbook Equation 9.17 and 9.18.
If the above conditions are violated, arbitrage opportunities exist.
K
80
85
90
T
0.271 2
0.271 2
0.271 2
Cb
6.5
3.2
1.2
www.actuary88.com
Ca
6.7
3.4
1.35
c
Yufeng
Guo
20
www.actuary88.com
c
Yufeng
Guo
21
www.actuary88.com
c
Yufeng
Guo
22
Chapter 10
4130 + Be0.08(0.5) = 25
4Su + BerT = Vu
rT
4Sd + Be = Vd
480 + Be0.08(0.5) = 0
B = 38. 431 6
4 = 0.5
So the option premium is:
V = 4S + B = 0.5 100 + (38. 431 6) = 11. 568 4
4130 + Be0.08(0.5) = 0
4Su + BerT = Vu
rT
4Sd + Be = Vd
480 + Be0.08(0.5) = 25
B = 62. 451 3
4 = 0.5
So the option premium is:
V = 4S + B = 0.5 100 + 62. 451 3 = 12. 451 3
Problem 10.2.
The stock price today is S = 100.The stock at T
Su = uS = 1.3 100 = 130
Sd = dS = 0.8 100 = 80
a. Payo at T is either
Vu = max (0, Su K) = max (0, 130 95) = 35
Vd = max (0, Sd K) = max (0, 80 95) = 0
We hold a replicating portfolio (4, B) at t = 0. This portfolio will have
value Vu if the stock goes up to Su or Vd if the stock goes down to Sd . We set
up the following equations:
4130 + Be0.08(0.5) = 35
4Su + BerT = Vu
rT
4Sd + Be = Vd
480 + Be0.08(0.5) = 0
B = 53. 804 2
4 = 0.7
c
Yufeng
Guo
24
4Su + Be = Vu
4130 + Be0.08(0.5) = 0
4Sd + BerT = Vd
480 + Be0.08(0.5) = 15
B = 37. 470 9
4 = 0.3
www.actuary88.com
c
Yufeng
Guo
25
c
Yufeng
Guo
26
Vd =?
Sdd = 100 0.82 = 64
(d , Bd )
Vdd = 0
We start from right to left.
Step 1
Calculate (u , Bu ), the replicating portfolio at Node u which will
produce
the
payo
Vud and Vdd .
rh
4u Sud + Bu e = Vud
4u 104 + Bu e0.08(0.5) = 9
Bu = 91. 2750
4u = 1
The premium at Node u is:
Vu = 4u Su + Bu = 1 (130) 91. 2750 = 38. 725
Step 2
Calculate(d , Bd ), the replicating portfolio at Node d which will
produce
the
payo
Vud and Vdd .
rh
4d Sdd + Bd e = Vdd
4d 64 + Bd e0.08(0.5) = 0
4d = 0.225 , Bd = 13. 835 4
The premium at Node d is:
Vd = 4d Sd + Bd = 0.225 (80) 13. 835 4 = 4. 164 6
Step 3
Calculate(, B), the replicating portfolio at time zero, which will
produce
the
payo
Vu and Vd .
4Su + BerT = Vu
4130 + Be0.08(0.5) = 38. 725
rT
4Sd + Be = Vd
480 + Be0.08(0.5) = 4. 164 6
4 = 0.691 2
B = 49. 127 1
The premium at time zero is:
V = 4S + B = 0.691 2 (100) 49. 127 1 = 19. 992 9
The final diagram is:
www.actuary88.com
c
Yufeng
Guo
27
S = 100
V = 19. 992 9
= 0.691 2
B = $49. 127 1
Sd = 100 (0.8) = 80
Vd = 4. 164 6
d = 0.225
Bd = $13. 835 4
Problem 10.5.
This question asks us to redo the previous problem by setting the initial
stock price to 80, 90, 110, 120, and 130.
S = 80.
Period 0
1
Su = 80 (1.3) = 104
Vu =?
(u , Bu ) =?
S = 80
V =?
(, B) =?
Sd = 80 (0.8) = 64
Vd =?
(d , Bd ) =?
4d 83. 2 + Bd e0.08(0.5) = 0
4d = 0.0
4d 51. 2 + Bd e0.08(0.5) = 0
Vd = 4d Sd + Bd = 0 (64) 0 = 0
www.actuary88.com
Bd = 0
4 = 0.465 058
c
Yufeng
Guo
B = 28. 596 7
28
www.actuary88.com
Sd = 80 (0.8) = 64
Vd = 0
d = 0
Bd = 0
c
Yufeng
Guo
29
1
Su = 90 (1.3) = 117
Vu =?
4u =?
Bu =?
S = 90
V =?
=?
B =?
Sd = 90 (0.8) = 72
Vd =?
d =?
Bd =?
4d 93. 6 + Bd e0.08(0.5) = 0
4d = 0.0
4d 57. 6 + Bd e0.08(0.5) = 0
Vd = 4d Sd + Bd = 0 (72) 0 = 0
Bd = 0
S = 90
V = 12. 226 52
4 = 0.587 161
B = 40. 617 97
www.actuary88.com
Su = 90 (1.3) = 117
Vu = 26. 422 26
4u = 0.976 068
Bu = 87. 777 7
Sd = 90 (0.8) = 72
Vd = 0
d = 0
Bd = 0
B = 40. 617 97
c
Yufeng
Guo
30
1
Su = 110 (1.3) = 143
Vu =?
4u =?
Bu =?
S = 110
V =?
=?
B =?
2
Suu = 110 1.32 = 185. 9
Vuu = 185. 9 95 = 90. 9
Bu = 91. 2750
S = 110
28. 406 15
4 = 0.777 235
B = 57. 089 7
www.actuary88.com
Sd = 110 (0.8) = 88
Vd = 8. 977 092
d = 0.440 909
Bd = 29. 822 9
Bd = 29. 822 9
B = 57. 089 7
c
Yufeng
Guo
31
1
Su = 120 (1.3) = 156
Vu =?
4u =?
Bu =?
S = 120
V =?
=?
B =?
Bu = 91. 275
S = 120
V = 36. 818 6
4 = 0.848 925
B = 65. 052 4
www.actuary88.com
B = 65. 052 4
2
Suu = 120 1.32 = 202. 8
Vuu = 202. 8 95 = 107. 8
Sd = 120 (0.8) = 96
Vd = 13. 789 528
4d = 0.620 833
Bd = 45. 810 44
c
Yufeng
Guo
32
1
Su = 130 (1.3) = 169
Vu =?
4u =?
Bu =?
S = 130
V =?
=?
B =?
Vd = 18. 602
Sdd = 130 0.82 = 83. 2
4d = 0.773 077
Vdd = 0
Bd = 61. 7980
Notice that the delta for a call is always positive. A positive delta means
buying stocks. If you sell a call, the risk you face is that the stock price may go
www.actuary88.com
c
Yufeng
Guo
33
www.actuary88.com
c
Yufeng
Guo
34
1
Su = 100 (1.3) = 130
Vu =?
4u =?
Bu =?
S = 100
V =?
=?
B =?
Sd = 100 (0.8) = 80
Vd =?
d =?
Bd =?
4u 169 + Bu e0.08(0.5) = 0
4u = 0
4u 104 + Bu e0.08(0.5) = 0
Vu = 4u Su + Bu = 0 (130) + 0 = 0
Bu = 0
4d 104 + Bd e0.08(0.5) = 0
Bd = 77. 439 6
4d = 0.775
4d 64 + Bd e0.08(0.5) = 31
Vd = 4d Sd + Bd = 0.775 (80) + 77. 439 6 = 15. 439 6
4130 + Be0.08(0.5) = 0
4 = 0.308 792
B = 38. 568 932
480 + Be0.08(0.5) = 15. 439 6
V = 4S + B = 0.308 792 (100) + 38. 568 932 = 7. 689 732
The final diagram is:
Period 0
1
2
Vd = 15. 439 6
Sdd = 100 0.82 = 64
4d = 0.775
Vdd = 95 64 = 31
Bd = 77. 439 6
www.actuary88.com
c
Yufeng
Guo
35
1
Su = 80 (1.3) = 104
Vu =?
4u =?
Bu =?
S = 80
V =?
=?
B =?
4u 135. 2 + Bu e0.08(0.5) = 0
4u = 0.226 923
4u 83. 2 + Bu e0.08(0.5) = 11. 8
Vu = 4u Su + Bu = 0.226 923 (104) + 29. 477 02 = 5. 877
Bu = 29. 477 02
Bd = 91. 275
Vd = 27. 275
Sdd = 80 0.82 = 51. 2
4d = 1
Vdd = 95 51. 2 = 43. 8
Bd = 91. 275
www.actuary88.com
c
Yufeng
Guo
36
1
Su = 90 (1.3) = 117
Vu =?
4u =?
Bu =?
S = 90
V =?
=?
B =?
4u 152. 1 + Bu e0.08(0.5) = 0
Bu = 3. 497 27
4u = 0.02 393
4u 93. 6 + Bu e0.08(0.5) = 1. 4
Vu = 4u Su + Bu = 0.02 393 (117) + 3. 497 27 = 0.697 5
4d 93. 6 + Bd e0.08(0.5) = 1. 4
4d = 1
4d 57. 6 + Bd e0.08(0.5) = 37. 4
Vd = 4d Sd + Bd = 1 (72) + 91. 275 = 19. 275
Bd = 91. 275
Vd = 19. 275
Sdd = 90 0.82 = 57. 6
4d = 1
Vdd = 95 57. 6 = 37. 4
Bd = 91. 275
www.actuary88.com
c
Yufeng
Guo
37
1
Su = 110 (1.3) = 143
Vu =?
4u =?
Bu =?
S = 110
V =?
=?
B =?
Sd = 110 (0.8) = 88
Vd =?
d =?
Bd =?
4u 185. 9 + Bu e0.08(0.5) = 0
4u = 0
4u 114. 4 + Bu e0.08(0.5) = 0
Vu = 4u Su + Bu = 0 (143) + 0 = 0
Bu = 0
4d 114. 4 + Bd e0.08(0.5) = 0
4d = 0.559 091
4d 70. 4 + Bd e0.08(0.5) = 24. 6
Vd = 4d Sd + Bd = 0.559 091 (88) + 61. 452 1 = 12. 252
Bd = 61. 452 1
4143 + Be0.08(0.5) = 0
4 = 0.222 76
B = 30. 606 14
488 + Be0.08(0.5) = 12. 252
V = 4S + B = 0.222 76 (110) + 30. 606 14 = 6. 103
The final diagram is:
Period 0
1
2
Vd = 12. 252
Sdd = 110 0.82 = 70. 4
4d = 0.559 091
Vdd = 95 70. 4 = 24. 6
Bd = 61. 452 1
www.actuary88.com
c
Yufeng
Guo
38
1
Su = 120 (1.3) = 156
Vu =?
4u =?
Bu =?
S = 120
V =?
=?
B =?
Sd = 120 (0.8) = 96
Vd =?
d =?
Bd =?
4u 202. 8 + Bu e0.08(0.5) = 0
4u = 0
4u 124. 8 + Bu e0.08(0.5) = 0
Vu = 4u Su + Bu = 0 (156) + 0 = 0
Bu = 0
4d 124. 8 + Bd e0.08(0.5) = 0
4d = 0.379 167
4d 76. 8 + Bd e0.08(0.5) = 18. 2
Vd = 4d Sd + Bd = 0.379 167 (96) + 45. 464 56 = 9. 065
Bd = 45. 464 56
4156 + Be0.08(0.5) = 0
4 = 0.151 083
B = 22. 644 85
496 + Be0.08(0.5) = 9. 065
V = 4S + B = 0.151 083 (120) + 22. 644 85 = 4. 514 89
The final diagram is:
Period 0
1
Su = 120 (1.3) = 156
Vu = 0
4u = 0
Bu = 0
S = 120
V = 4. 514 89
4 = 0.151 083
B = 22. 644 85
Sd = 120 (0.8) = 96
Vd = 9. 065
4d = 0.379 167
Bd = 45. 464 56
www.actuary88.com
c
Yufeng
Guo
39
1
Su = 130 (1.3) = 169
Vu =?
4u =?
Bu =?
S = 130
V =?
=?
B =?
4u 219. 7 + Bu e0.08(0.5) = 0
4u = 0
4u 135. 2 + Bu e0.08(0.5) = 0
Vu = 4u Su + Bu = 0 (169) + 0 = 0
Bu = 0
4d 135. 2 + Bd e0.08(0.5) = 0
4d = 0.226 923
4d 83. 2 + Bd e0.08(0.5) = 11. 8
Vd = 4d Sd + Bd = 0.226 923 (104) + 29. 477 0 = 5. 877
Bd = 29. 477 0
4169 + Be0.08(0.5) = 0
4 = 0.09 041 5
B = 14. 681 05
4104 + Be0.08(0.5) = 5. 877
V = 4S + B = 0.09 041 5 (130) + 14. 681 05 = 2. 927 1
The final diagram is:
Period 0
1
2
Vd = 5. 877
Sdd = 130 0.82 = 83. 2
4d = 0.226 923
Vdd = 95 83. 2 = 11. 8
Bd = 29. 477 0
www.actuary88.com
c
Yufeng
Guo
40
1
Su = 100 (1.3) = 130
EVu = 0
Vu =?
4u =?
Bu =?
S = 100
EV = 0
V =?
=?
B =?
Sd = 100 (0.8) = 80
EVd = 95 80 = 15
Vd =?
d =?
Bd =?
4u 169 + Bu e0.08(0.5) = 0
Bu = 0
4u = 0
4u 104 + Bu e0.08(0.5) = 0
Vu = max (4u Su + Bu , EVu ) = max [0 (130) + 0, 0] = 0
This is the logic behind the calculation of Vu . The American put can be
exercised at the end of Period 1 and end of Period 2.
If exercised at the end of Period 1, the put is worth EVu = max (0, 95 130) =
0 at the end of Period 1.
If exercised at Period 2, the put is worth 4u Su + Bu = 0 (130) + 0 = 0 at
the end of Period 1.
www.actuary88.com
c
Yufeng
Guo
41
4d 104 + Bd e0.08(0.5) = 0
Bd = 77. 439 6
4d = 0.775
4d 64 + Bd e0.08(0.5) = 31
Vd = max (4d Sd + Bd , EVd ) = max [0.775 (80) + 77. 439 6, 15] = 15. 439 6
4130 + Be0.08(0.5) = 0
4 = 0.308 792
B = 38. 568 932
480 + Be0.08(0.5) = 15. 439 6
V = max (4S + B, EV ) = max [0.308 792 (100) + 38. 568 932 , 0] = 7. 689 732
Sd = 100 (0.8) = 80
EVd = 95 80 = 15
Vd = 15. 439 6
4d = 0.775
Bd = 77. 439 6
Problem 10.9.
a.
time 0
T
Su = 100 (1.2) = 120
Vu = 120 50 = 70
S = 100
V =?
=?, B =?
c
Yufeng
Guo
42
4120 + Be0.07696(1) = 70
4=1
B = 46. 296 3
4105 + Be0.07696(1) = 55
V = 4S + B = 1 (100) 46. 296 3 = 53. 703 7
The final diagram is:
time 0
S = 100
V = 53. 703 7
4 = 1, B = 46. 296 3
T
Su = 100 (1.2) = 120
Vu = 120 50 = 70
b. Now the stock price has a bigger increase and a bigger decrease. We
are not clear whether the call premium will increase or decrease. We have to
calculate the premium.
time 0
T
Su = 100 (1.4) = 140
Vu = 140 50 = 90
S = 100
V =?
=?, B =?
Sd = 100 (0.6) = 60
Vd = 60 50 = 10
4140 + Be0.07696(1) = 90
4=1
B = 46. 296 3
460 + Be0.07696(1) = 10
V = 4S + B = 1 (100) 46. 296 3 = 53. 703 7
The call premium is the same. Whats going on?
It turns out that as long Su > Sd > K, the call premium is fixed regardless
of how you choose u and d.
time 0
T
Su = uS > K
Vu = uS K
S
V =?
=?, B =?
Sd = dS > K
Vd = dS K
4uS + Berh = uS K
4dS + Berh = dS K
www.actuary88.com
c
Yufeng
Guo
43
C = 4S + B = S Kerh = 100
We can explain this using the putcall parity. If Su > Sd > K, the put is
never exercised.
C + P V (K) = P + S
Since P = 0, we have C = S P V (K) = S Kerh = 10050e0.07696(1) =
53. 703 7
c. I think the question wants us to compare b and c, not a and c.
Compared with b, now u is the same but d gets smaller. In fact, now at
Node d, the call expires worthless.
time 0
S = 100
V =?
=?, B =?
T
Su = 100 (1.4) = 140
Vu = 140 50 = 90
Sd = 100 (0.4) = 40
Vd = 0
We might think the call option should be worth less since it expires worthless
at Node
d.
4140 + Be0.07696(1) = 90
4 = 0.9
B = 33. 333
440 + Be0.07696(1) = 0
V = 4S + B = 0.9 (100) 33. 333 = 56. 667 > 53. 703 7
Why did the call premium goes up to 56. 667? Does this lead to arbitrage?
First, Im going to show you this is not an arbitrage. Now we have two call
options:
Call b
Call c
time 0
T
time 0
T
Su = 100 (1.4) = 140
Su = 100 (1.4) = 140
Vu = 140 50 = 90
Vu = 140 50 = 90
S = 100
S = 100
V = 53. 703 7
V = 56. 667
4 = 1, B = 46. 296 3
4 = 0.9, B = 33. 333
Sd = 100 (0.6) = 60
Sd = 100 (0.4) = 40
Vd = 60 50 = 10
Vd = 0
What if we buy Call b and simultaneously sell Call c? Does this lead to
arbitrage?
The answer is No because these two calls are on two dierent stocks. If
these two options have the same underlying asset, then well make free money
www.actuary88.com
c
Yufeng
Guo
44
Problem 10.10.
h = T /n = 1/3
u = e(r)h+
d = e(r)h
u =
e(r)h d
e(0.080)1/3 0.863 693
=
= 0.456 806
ud
1. 221 246 0.863 693
c
Yufeng
Guo
45
R
, EVud = max (15. 040 33, 10. 478 2) = 15. 040 33
Vud = max Vud
R
= erh ( u Vudd + d Vddd ) = e(0.08)1/3 (0.456 806 0 + 0.543 194 0) =
Vdd
Now we have:
Period 1
VuR = erh (u Vuu + d Vud ) = e(0.08)1/3 (0.456 806 56. 644 02 + 0.543 194 15. 040 33) =
33. 149 27
Vu = max VuR , EVu = max (33. 149 27, 27. 124 6) = 33. 149 27
VdR = erh (u Vud + d Vdd ) = e(0.08)1/3 (0.456 806 15. 040 33 + 0.543 194 0) =
6. 689 72
S = 100
EV = 100 95 = 5
V =?
1
Su = 100 (1. 221 246) = 122. 124 6
EVu = 122. 124 6 95 = 27. 124 6
Vu = 33. 149 27
V R = erh (u Vu + d Vd ) = e(0.08)1/3 (0.456 806 33. 149 27 + 0.543 194 6. 689 72) =
18. 282 51
c
Yufeng
Guo
46
Period 2
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.08)1/3 (0.456 806 0 + 0.543 194 0) =
R
= erh ( u Vuud + d Vudd ) = e(0.08)1/3 (0.456 806 0 + 0.543 194 3. 899 2) =
Vud
2. 062 29
R
R
, EVdd = max (17. 903 58, 20. 403 4) = 20. 403 4
Vdd = max Vdd
At the dd node, the exercise value is greater than the rollback value. The
put is exercised at dd. Now we have:
www.actuary88.com
c
Yufeng
Guo
47
VuR = erh (u Vuu + d Vud ) = e(0.08)1/3 (0.456 806 0 + 0.543 194 2. 062 29) =
1. 090 75
Vd = max VdR , EVd = max (11. 708 64, 8. 630 7) = 11. 708 64
Now we have:
Period 0 1
Su = 100 (1. 221 246) = 122. 124 6
EVu = 0
Vu = 1. 090 75
S = 100
EV = 0
V =?
Sd = 100 (0.863 693) = 86. 369 3
EVd = 95 86. 369 3 = 8. 630 7
Vd = 11. 708 64
V R = erh (u Vu + d Vd ) = e(0.08)1/3 (0.456 806 1. 090 75 + 0.543 194 11. 708 64) =
6. 677 85
c
Yufeng
Guo
48
3
Vuuu = 0
Vuu =?
Vuud = 0
Vud =?
Vudd = 3. 899 2
Vdd =?
rh
Vuu = e
0
Vud = erh ( u Vuud + d Vudd ) = e(0.08)1/3 (0.456 806 0 + 0.543 194 3. 899 2) =
2. 062 29
Vdd = erh (u Vudd + d Vddd ) = e(0.08)1/3 (0.456 806 3. 899 2 + 0.543 194 30. 571 5) =
17. 903 58
Now we have:
Period 1 2
Vuu = 0
Vu =?
Vud = 2. 062 29
Vd =?
Vdd = 17. 903 58
Vu = erh (u Vuu + d Vud ) = e(0.08)1/3 (0.456 806 0 + 0.543 194 2. 062 29) =
1. 090 75
Vd = erh ( u Vud + d Vdd ) = e(0.08)1/3 (0.456 806 2. 062 29 + 0.543 194 17. 903 58) =
10. 386 48
Now we have:
Period 0 1
Vu = 1. 090 75
V =?
Vd = 10. 386 48
rh
V =e
( u Vu + d Vd ) = e(0.08)1/3 (0.456 806 1. 090 75 + 0.543 194 10. 386 48) =
5. 978 6
Check whether the putcall parityholds:
C + KerT = 18. 282 51 + 95e(0.08)1 = 105. 978 6
P + S = 5. 978 6 + 100 = 105. 978 6
C + KerT = P + S
Problem 10.11.
h = T /n = 1/3
u = e(r)h+
www.actuary88.com
= e(0.080.08)1/3+0.3
1/3
= 1. 189 11
c
Yufeng
Guo
49
u =
e(r)h d
e(0.080.08)1/3 0.840 965
=
= 0.456 807
ud
1. 189 11 0.840 965
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.08)1/3 (0.456 807 73. 138 1 + 0.543 193 23. 911) =
45. 177 3
R
R
= erh (u Vuud + d Vudd ) = e(0.08)1/3 (0.456 807 23. 911 + 0.543 193 0) =
Vud
10. 635 3
R
R
Vdd
= erh ( u Vudd + d Vddd ) = e(0.08)1/3 (0.456 807 0 + 0.543 193 0) =
www.actuary88.com
c
Yufeng
Guo
50
2
Vuu = 46. 398 3
Vu = max VuR , EVu = max (26. 262 3, 23. 911) = 26. 262 3
VdR = erh ( u Vud + d Vdd ) = e(0.08)1/3 (0.456 807 10. 635 3 + 0.543 193 0) =
4. 730 4
1
Vu = 26. 262 3
S = 100
EV = 100 95 = 5
Vu =?
Vd = 4. 730 4
V R = erh ( u Vu + d Vd ) = e(0.08)1/3 (0.456 807 26. 262 3 + 0.543 193 4. 730 4) =
14. 183 0
Vuu
Vu
V
Vud
Vd
Vdd
www.actuary88.com
51
R
= erh (u Vuud + d Vudd ) = e(0.08)1/3 (0.456 807 0 + 0.543 193 10. 903 5) =
Vud
5. 766 9
R
R
= erh ( u Vudd + d Vddd ) = e(0.08)1/3 (0.456 807 10. 903 5 + 0.543 193 35. 525 1) =
Vdd
23. 638 9
R
www.actuary88.com
c
Yufeng
Guo
52
2
Vuu = 0
VdR = erh ( u Vud + d Vdd ) = e(0.08)1/3 (0.456 807 5. 766 9 + 0.543 193 24. 277 8) =
15. 405 6
Vd = max VdR , EVd = max (15. 405 6, 10. 903 5) = 15. 405 6
Period 0
1
Vu = 3. 050 1
S = 100
EV = 0
Vu =?
Vd = 15. 405 6
V R = erh ( u Vu + d Vd ) = e(0.08)1/3 (0.456 807 3. 050 1 + 0.543 193 15. 405 6) =
9. 504 7
Vu
Suud = 100 1. 189 112 (0.840 965) = 118. 9110
Vuud = 0
V
Vud
c
Yufeng
Guo
53
Vud = erh (u Vuud + d Vudd ) = e(0.08)1/3 (0.456 807 0 + 0.543 193 10. 903 5) =
5. 766 9
Vdd = erh ( u Vudd + d Vddd ) = e(0.08)1/3 (0.456 807 10. 903 5 + 0.543 193 35. 525 1) =
23. 638 9
Vu = erh (u Vuu + d Vud ) = e(0.08)1/3 (0.456 807 0 + 0.543 193 5. 766 9) =
3. 050 1
Vd = erh ( u Vud + d Vdd ) = e(0.08)1/3 (0.456 807 5. 766 9 + 0.543 193 23. 638 9) =
15. 067 6
V = erh (u Vu + d Vd ) = e(0.08)1/3 (0.456 807 3. 050 1 + 0.543 193 15. 067 6) =
9. 325 9
So the European put is worth 9. 325 9.
Verify the putcall parity:
C + KerT = 13. 941 5 + 95e(0.08)1 = 101. 637 55
P + SeT = 9. 325 9 + 100e(0.08)1 = 101. 637 53
Ignoring rounding dierence, we get C + KerT = P + SeT
The putcall parity holds.
Problem 10.12.
a. h = T /n =0.5/2 = 0.25
0.25
= e(0.08)0.250.3
= 0.878 095
e(r)h d
e(0.08)0.25 0.878 095
=
= 0.462 57
ud
1. 185 305 0.878 095
d = 1 u = 1 0.462 57 = 0.537 43
b. Since the stock doesnt pay dividend, the American call and an otherwise
identical European call have the same value.
Period 0 1
2
www.actuary88.com
54
P
V = erT payof f RiskN eutralP rob
= e0.08(0.5) 16. 197 92 0.462 572 + 1. 632 4 2 0.462 57 0.537 43 + 0 0.537 432
= 4. 109 8
Please note that the above shortcut works only for European options.
c. Even if the stock doesnt pay dividend, it may be still optimal to exercise
an American put early.
Calculate the premium of the American put
Period 0
1
Su = 40 (1. 185 305) = 47. 412 2
EVu = 0
Vu =?
S = 40
EV = 0
V =?
VdR = erh ( u Vud + d Vdd ) = e(0.08)1/3 (0.456 807 0 + 0.543 193 9. 157 97) =
4. 843 6
c
Yufeng
Guo
55
Vu
Vd
S = 40
V = 4. 109 7
(, B)
c
Yufeng
Guo
56
4Su + Berh = Vu
4Sd + Berh = Vd
4 = 0.607 41
B = 20. 186 7
207 9
4u = 1
Bu = 39.
4d = 0.151 28
Bd = 4.
573 5
At t = 0, we
c
Yufeng
Guo
57
Suppose the stock price goes up to Su and we liquidate our position at Node
u. Well
1. Sell our 4 = 0.607 41 share of the stock, receiving 4Su = 0.607 41 (47. 412 2) =
$28. 798 6
2. Pay the bank Berh = 20. 186 7e0.08(0.25) = $20. 594 5
3. Buy the call from the open market for the fair price of Vu = $8. 204 2
The net receipt is: 28. 798 6 20. 594 5 8. 204 2 = 0.000 1 = $0 (if we
ignore rounding errors)
So we receive $0.890 3 free money at t = 0 without incurring any liability at
time h.
Suppose the stock price goes up to Sd and we liquidate our position at Node
u. Well
1. Sell our 4 = 0.607 41 share of the stock, receiving 4Sd = 0.607 41 (35. 123 8) =
$21. 334 5
2. Pay the bank Berh = 20. 186 7e0.08(0.25) = $20. 594 5
3. Buy the call from the open market for the fair price of Vd = $0.740 1
The net receipt is: 21. 334 5 20. 594 5 0.740 1 = 0.000 1 = 0 (if we ignore
rounding errors)
So we receive $0.890 3 free money at t = 0 without incurring any liability at
time h.
c
Yufeng
Guo
58
1. Our written call is exercised against us. We need to pay the call holder
Vuu = 56. 197 92 40 = 16. 197 92.
2. We sell u stocks in the market and pay o our loan from the bank. Our
net cash receipt is u Suu + Bu erh = 1 56. 197 92 39. 207 9e0.08(0.25) =
16. 197 97.
3. Our net cash flow is zero (ignoring rounding).
If Sud = 40 (1. 185 305) (0.878 095) = 41. 632 4
1. Our written call is exercised against us. We need to pay the call holder
Vud = 41. 632 4 40 = 1. 632 4
2. We sell u stocks in the market and pay o our loan from the bank. Our
net cash receipt is u Sud + Bu erh = 1 41. 632 4 39. 207 9e0.08(0.25) =
1. 632 4
3. Our net cash flow is zero.
So we receive $0.890 3 free money at t = 0 without incurring any liability at
T.
Situation #2 the stock price goes down to Sd = 40 (0.878 095) = 35.
123 8 at time h
Well change our replicating portfolio from (, B) = (0.607 41, B = $20. 186 7)
to (d , Bd ) = (0.151 28, $4. 573 5) at h. We need to sell 0.607 4 0.151 28 =
0.456 12 share of the stock, receiving 0.456 12 (35. 123 8) = $16. 020 7. We immediately send a check of $16. 020 7 to the bank to partially pay our debt. Now
our remaining debt to the bank is
20. 186 7e0.08(0.25) 16. 020 7 = 4. 573 8 = 4. 573 5 (ignore rounding)
Now at time h, our replicating portfolio is exactly (d , Bd ) = (0.151 28, $4. 573 5)
Then at T = 2h
If Sdu = 40 (1. 185 305) (0.878 095) = 41. 632 4
1. Our written call is exercised against us. We need to pay the call holder
Vdu = 41. 632 4 40 = 1. 632 4
2. We sell d stocks in the market and pay o our loan from the bank. Our
net cash receipt is d Sdu +Bd erh = 0.151 2841. 632 44. 573 5e0.08(0.25) =
1. 632 3
3. Our net cash flow is zero (ignoring rounding).
www.actuary88.com
c
Yufeng
Guo
59
Problem 10.14.
We can use the general binomial tree formula by setting S = 0.92 and
= r = 0.03.
h = T /n = 0.75/3 = 0.25
u = 1.2
d = 0.9
e(r)h d
e(0.040.03)0.25 0.9
u =
=
= 0.341 677
ud
1.2 0.9
d = 1 u = 1 0.341 677 = 0.658 323
Well calculate part b first.
b. Calculate the American call premium.
www.actuary88.com
c
Yufeng
Guo
60
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.04)0.25 (0.341 677 0.739 76 + 0.658 323 0.342 32) =
0.473 359
R
2
Vuu = 0.474 8
c
Yufeng
Guo
61
1
Vu = 0.254 882
S = 0.92
EV = 0.92 0.85 = 0.07
V =?
Vd = 0.058 68
V R = erh (u Vu + d Vu ) = e(0.04)0.25 (0.341 677 0.254 882 + 0.658 323 0.058 68) =
0.124 467
Vud
Vd
Vdd
Vuu = erh ( u Vuuu + d Vuud ) = e(0.04)0.25 (0.341 677 0.739 76 + 0.658 323 0.342 32) =
0.473 359
Vud = erh (u Vuud + d Vudd ) = e(0.04)0.25 (0.341 677 0.342 32 + 0.658 323 0.044 24) =
0.144 633
Vdd = erh ( u Vudd + d Vddd ) = e(0.04)0.25 (0.341 677 0.044 24 + 0.658 323 0) =
0.01 496 5
Vu = erh (u Vuu + d Vud ) = e(0.04)0.25 (0.341 677 0.473 359 + 0.658 323 0.144 633) =
0.254 394
VdR = erh (u Vud + d Vdd ) = e(0.04)0.25 (0.341 677 0.144 633 + 0.658 323 0.01 496 5) =
0.05 8680
V = erh (u Vu + d Vu ) = e(0.04)0.25 (0.341 677 0.254 394 + 0.658 323 0.058 68) =
0.124 302
So the European call premium is 0.124 302 .
www.actuary88.com
c
Yufeng
Guo
62
u = 1.2
d = 0.9
e(r)h d
e(0.040.03)0.25 0.9
u =
=
= 0.341 677
ud
1.2 0.9
d = 1 u = 1 0.341 677 = 0.658 323
b. Calculate the American put premium.
Period 2
3
Vuu =?
Suud = 0.92 1. 22 (0.9) = 1. 192 32
Vuud = 0
Sud = 0.92 (1. 2) (0.9) = 0.993 6
EVud = 1 0.993 6 = 0.006 4
Vud =?
2
S
=
0.92
(1.
2)
0.9 = 0.894 24
udd
Vdd =?
Sddd = 0.92 0.93 = 0.670 68
Vddd = 1 0.670 68 = 0.329 32
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.04)0.25 (0.341 677 0 + 0.658 323 0) =
R
= erh ( u Vuud + d Vudd ) = e(0.04)0.25 (0.341 677 0 + 0.658 323 0.105 76) =
Vud
0.06893 1
R
= erh (u Vudd + d Vddd ) = e(0.04)0.25 (0.341 677 0.105 76 + 0.658 323 0.329 32) =
Vdd
0.250 418
R
www.actuary88.com
c
Yufeng
Guo
63
2
Vuu = 0
1
Vu = 0.04 492 7
S = 0.92
EV = 1 0.92 = 0.08
V =?
Vd = 0.189 389
V R = erh (u Vu + d Vu ) = e(0.04)0.25 (0.341 677 0.04 492 7 + 0.658 323 0.189 389) =
0.138 636
Vu
Suud = 0.92 1. 22 (0.9) = 1. 192 32
Vuud = 0
V
Vud
c
Yufeng
Guo
64
c
Yufeng
Guo
65
e(r)h d
e(0.010.05)1/3 0.931 398
=
= 0.485 57
ud
1. 045 402 0.931 398
d = 1 u = 1 0.485 57 = 0.514 43
Vuu =?
Suud = 120 1. 045 4022 (0.931 398) = 122. 147 1
Vuud = 122. 147 1 120 = 2. 147 1
Sud = 120 (1. 045 402) (0.931 398) = 116. 842 2
EVud = 0
Vud =?
Vdd =?
Sddd = 120 0.931 3983 = 96. 958 8
Vddd = 0
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.01)1/3 (0.485 57 17. 098 + 0.514 43 2. 147 1) =
9. 375 5
R
R
= erh (u Vudd + d Vddd ) = e(0.01)1/3 (0.485 57 0 + 0.514 43 0) =
Vdd
2
Vuu = 11. 143 8
c
Yufeng
Guo
66
1
Vu = 5. 925 9
S = 120
EV = 0
V =?
Vd = 0.502 9
V R = erh ( u Vu + d Vu ) = e(0.01)1/3 (0.485 57 5. 925 9 + 0.514 43 0.502 9) =
3. 125 7
Vu
Suud = 120 1. 045 4022 (0.931 398) = 122. 147 1
Vuud = 122. 147 1 120 = 2. 147 1
V
Vud
c
Yufeng
Guo
67
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.01)1/3 (0.485 57 0 + 0.514 43 0) =
R
= erh (u Vuud + d Vudd ) = e(0.01)1/3 (0.485 57 0 + 0.514 43 11. 173 4) =
Vud
5. 728 8
R
R
= erh ( u Vudd + d Vddd ) = e(0.01)1/3 (0.485 57 11. 173 4 + 0.514 43 23. 041 2) =
Vdd
17. 221 1
R
Period 1
2
Vuu = 0
c
Yufeng
Guo
68
VdR = erh ( u Vud + d Vdd ) = e(0.01)1/3 (0.485 57 5. 728 8 + 0.514 43 17. 221 1) =
11. 602 0
1
Vu = 2. 937 3
S = 120
EV = 0
V =?
Vd = 11. 602
V R = erh ( u Vu + d Vu ) = e(0.01)1/3 (0.485 57 2. 937 3 + 0.514 43 11. 602) =
7. 370 1
Since the exercise value is never greater than the rollback value, the American put is not early exercised. Hence the American put and the European put
have the same value.
Both the American put and European put are worth 7. 370 1 Yen.
c. Since the underlying asset generates dividend (the Yen interest rate is
like the dividend rate), it may be optimal to early exercise an American call.
However, even if the asset generates dividend, it may not be optimal to early
exercise an American put.
Problem 10.17.
Method 1
Apply the stock binomial formula to futures
As explained in my study guide, we can apply the stock binomial formula
to options on futures if we do the following three things:
Set the dividend yield equal to the risk free rate (i.e. = r)
Cu Cd
Cu Cd
instead of 4 = erh
Su Sd
Su Sd
1d
u1
Su Cd Sd Cu
B = V = erh Cu
.
+ Cd
instead of B = erh
ud
ud
Su Sd
4=
www.actuary88.com
c
Yufeng
Guo
69
S = 300
EV = 300 290 = 10
V =?
(, B) =?
1
Su = 300 (1. 105 170 9) = 331. 551 27
Vu = 331. 551 27 290 = 41. 551 27
4=
h
= e0.1 1 = 1. 105 170 9
u = e
d = e h = e0.1 1 = 0.904 837 4
Next, we build a futures price tree.
Period 0
1
u
F1,1
= 300 (1. 105 170 9) = 331. 551 27
Vu = 331. 551 27 290 = 41. 551 27
F0,1 = 300
EV = 300 290 = 10
V =?
(, B) =?
d
F1,1
= 300 (0.904 837 4) = 271. 451 22
Vd = 0
www.actuary88.com
c
Yufeng
Guo
70
u
F0,1 = F0,1 (u 1) in the u node
F1,1
d
F1,1
F0,1 = F0,1 (d 1) in the d node
u
F0,1 + BerT = Vu
F1,1
F0,1 (u 1) + BerT = Vu
d
rT
F0,1 (d 1) + BerT = Vd
F1,1 F0,1 + Be = Vd
Since it costs nothing to enter a future contract, the call premium is equal
to B:
V = B = 18. 588 29
Problem 10.18.
Well apply the stock binomial formula to futures. We need to do the following three things:
Set the dividend yield equal to the risk free rate (i.e. = r)
Cu Cd
Cu Cd
instead of 4 = erh
Su Sd
Su Sd
1d
u1
Su Cd Sd Cu
B = V = erh Cu
.
+ Cd
instead of B = erh
ud
ud
Su Sd
4=
www.actuary88.com
c
Yufeng
Guo
71
3
Suuu = 1000u3 = 1681. 380 8
Vuuu = 1681. 380 8 1000 = 681. 380 8
Suud = 1000u2 d = 1000u = 1189. 1100
Vuud = 1189. 11 1000 = 189. 11
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.05)1/3 (0.456 807 681. 380 8 + 0.543 193 189. 11) =
407. 140 2
R
R
= erh ( u Vudd + d Vddd ) = e(0.05)1/3 (0.456 807 0 + 0.543 193 0) =
Vdd
www.actuary88.com
c
Yufeng
Guo
72
2
Vuu = 413. 982 6
Vu = max VuR , EVu = max (231. 370 7 , 189. 11) = 231. 370 7
VdR = erh ( u Vud + d Vdd ) = e(0.05)1/3 (0.456 807 84. 958 9 + 0.543 193 0) =
38. 168 4
1
Vu = 231. 370 7
S = 1000
EV = 0
V =?
Vd = 38. 168 4
V R = erh ( u Vu + d Vu ) = e(0.05)1/3 (0.456 807 231. 370 7 + 0.543 193 38. 168 4) =
124. 334 9
Vuu
Vu
V
Vud
Vd
Vdd
www.actuary88.com
3
Suuu = 1000u3 = 1681. 380 8
Vuuu = 1681. 380 8 1000 = 681. 380 8
Suud = 1000u2 d = 1000u = 1189. 1100
Vuud = 1189. 11 1000 = 189. 11
Sudd = 1000ud2 = 1000d = 840. 965
Vudd = 0
73
B = V = $122. 953 9
Vu Vd
228. 296 7 38. 168 4
4=
=
= 0.546 118
Su Sd
1000 (1. 189 11) 1000 (0.840 965 )
calculate the American put premium
Period 2
3
Suuu = 1000u3 = 1681. 380 8
Vuuu = 0
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.05)1/3 (0.456 807 0 + 0.543 193 0) =
c
Yufeng
Guo
74
Period 1
2
Vuu = 0
Vd = max VdR , EVd = max (194. 574 6 , 159. 035) = 194. 574 6
Period 0
1
Vu = 45. 386 3
S = 1000
EV = 0
V =?
Vd = 194. 574 6
V R = erh ( u Vu + d Vu ) = e(0.05)1/3 (0.456 807 45. 386 3 + 0.543 193 194. 574 6) =
124. 334 7
c
Yufeng
Guo
75
3
Suuu = 1000u3 = 1681. 380 8
Vuuu = 0
Vuu
Suud = 1000u2 d = 1000u = 1189. 1100
Vuud = 0
Vu
V
Vud
Vd
Vdd
Vud = erh (u Vuud + d Vudd ) = e(0.05)1/3 (0.456 807 0 + 0.543 193 159. 035) =
84. 958 9
Vdd = erh ( u Vudd + d Vddd ) = e(0.05)1/3 (0.456 807 159. 035 + 0.543 193 405. 251) =
287. 938 62
Vu = erh (u Vuu + d Vud ) = e(0.05)1/3 (0.456 807 0 + 0.543 193 84. 958 9) =
45. 386 3
Vd = erh ( u Vud + d Vdd ) = e(0.05)1/3 (0.456 807 84. 958 9 + 0.543 193 287. 938 62) =
191. 989 4
V = erh (u Vu + d Vu ) = e(0.05)1/3 (0.456 807 45. 386 3 + 0.543 193 191. 989 4) =
122. 9537
So the European call premium is $122. 9537.
Time zero replicating portfolio for the European call
B = V = $122. 9537
45. 386 3 191. 989 4
Vu Vd
=
= 0.421 1
4=
Su Sd
1000 (1. 189 11) 1000 (0.840 965 )
So at t = 0 we need to enter 0.421 1 futures contract as a seller and deposit
122. 9537 in a savings account.
Why the European call and a European put have the same premium
The standard putcall parity is:
C + P V (K) = P + S
S is the price of the underlying asset at time zero. For the futures contract,
S is the present value of the forward price:
S = P V (F0,T )
In this problem, the forward price and the strike price are equal (i.e. K =
F0,T ). Hence P V (K) = P V (F0,T ) = S. This gives us C = P .
www.actuary88.com
c
Yufeng
Guo
76
(r)h+ h
(0.050.03)1+0.3 1
u=e
= 1. 377 128
=e
e(r)h d
e(0.050.03)1 0.755 784
=
= 0.425 557
ud
1. 377 128 0.755 784
Vu
Suud = 100 1. 377 1282 (0.755 784) = 143. 333 0
Vuud = 143. 333 0 95 = 48. 3330
V
Vud
c
Yufeng
Guo
77
Vuu
Vu
V
Vud
Vd
Vdd
Vuu = erh ( u Vuuu + d Vuud ) = e(0.05)1 (0.425 557 166. 169 8 + 0.574 443 48. 333) =
Vud = erh (u Vuud + d Vudd ) = e(0.05)1 (0.425 557 0 + 0.574 443 16. 337 1) =
8. 927 0
Vdd = erh ( u Vudd + d Vddd ) = e(0.05)1 (0.425 557 16. 337 1 + 0.574 443 51. 828 9) =
34. 934 0
Vu = erh (u Vuu + d Vud ) = e(0.05)1 (0.425 557 0 + 0.574 443 8. 927 0) =
4. 8780
Vd = erh ( u Vud + d Vdd ) = e(0.05)1 (0.425 557 8. 927 0 + 0.574 443 34. 934 0) =
22. 702 6
V = erh (u Vu + d Vu ) = e(0.05)1 (0.425 557 4. 8780 + 0.574 443 22. 702 6) =
14. 379 9
So the European put premium is $14. 379 9.
c. If we switch S and K and switch r and and recalculate the option price,
what happens?
After the switch, we have:
S = 95
K = 100
r = 3%
= 5%
e(r)h d
e(0.030.05)1 0.726 149
=
= 0.425 557
ud
1. 323 13 0.726 149
c
Yufeng
Guo
78
Vuu
Vu
V
Vud
Vd
Vdd
Vuu = erh ( u Vuuu + d Vuud ) = e(0.03)1 (0.425 557 120. 055 + 0.574 443 20. 768 7) =
61. 158 1
Vud = erh ( u Vuud + d Vudd ) = e(0.03)1 (0.425 557 20. 768 7 + 0.574 443 0) =
8. 577 1
Vdd = erh ( u Vudd + d Vddd ) = e(0.03)1 (0.425 557 0 + 0.574 443 0) =
0
Vu = erh (u Vuu + d Vud ) = e(0.03)1 (0.425 557 61. 158 1 + 0.574 443 8. 577 1) =
30. 038 5
Vd = erh ( u Vud + d Vdd ) = e(0.03)1 (0.425 557 8. 577 1 + 0.574 443 0) =
3. 542 2
V = erh ( u Vu + d Vd ) = e(0.03)1 (0.425 557 30. 038 5 + 0.574 443 3. 542 2) =
14. 3799
The call premium after the switch is equal to the put premium before the
switch.
Calculate the European put premium
Period 0
Vuu
Vu
V
Vud
Vd
Vdd
www.actuary88.com
79
Vud = erh (u Vuud + d Vudd ) = e(0.03)1 (0.425 557 0 + 0.574 443 33. 720 7) =
18. 798 1
Vdd = erh ( u Vudd + d Vddd ) = e(0.03)1 (0.425 557 33. 720 7 + 0.574 443 63. 625 2) =
49. 394 8
Vu = erh (u Vuu + d Vud ) = e(0.03)1 (0.425 557 0 + 0.574 443 18. 798 1) =
10. 479 3
Vd = erh ( u Vud + d Vdd ) = e(0.03)1 (0.425 557 18. 798 1 + 0.574 443 49. 394 8) =
35. 299 1
V = erh (u Vu + d Vd ) = e(0.03)1 (0.425 557 10. 479 3 + 0.574 443 35. 299 1) =
24. 005 8
The put premium after the switch is equal to the call premium before the
switch.
By the way, you can also use the textbooks spreadsheet "optbasics2" to
calculate the European option premium and verify
the European call price is $14. 379 9 (which is the European put price
before the switch)
the European put price is $24. 005 8 (which is the European call price
before the switch)
What a coincidence, you might wonder. Why? This can be explained using
the BlackScholes option formulas:
The price of a European call option is:
C = SeT N (d1 ) KerT N (d2 )
(Textbook 12.1)
(Textbook 12.3)
S
1 2
SeT
SeT
1
ln
+ r+ T
ln
ln
+ 2T
rT
rT
1
K
2
Ke 2
Ke
=
=
+ T
d1 =
2
T
T
T
(Textbook 12.2a)
d2 = d1 T
(Textbook 12.2b)
c
Yufeng
Guo
80
d1 =
+ T
2
T
SeT
SeT
ln
ln
rT
rT
1
1
Ke
Ke
+ T T =
T
d2 = d1 T =
2
2
T
T
SeT
SeT
ln KerT
ln KerT
1
1
+ T
T
KerT N
C = SeT N
T
2
2
T
SeT
SeT
ln KerT
ln
1
1
rT
SeT N Ke
P = KerT N
+
T
2
2
T
T
ln
r
P = KerT N (d2 ) SeT N (d1 ) KS = SeT N (d2 )KerT N (d1 )
r
SeT
KerT
ln
ln KerT
T
1
1
Se
+ T
+ T
=
d1 =
T
2
2
T
KS
KerT
SeT
ln
ln
T
rT
1
1
Ke
T =
T
d1 = Se
2
2
T
T
KerT
KerT
ln
ln
T
T
1
1
Se
Se
+ T T =
T
d2 = d1 T =
2
2
T
T
KerT
KerT
SeT
ln
ln
ln
T
rT
1
SeT + 1 T =
Ke
+ T =
+
d2 = Se
2
2
T
T
T
1
T
2
KerT
KerT
ln
ln
SeT
SeT
1
1
+ T
T
SeT N
C = KerT N
T
2
2
T
SeT
SeT
ln KerT
ln KerT
1
1
P = SeT N
+ T
T
KerT N
T
2
2
T
www.actuary88.com
c
Yufeng
Guo
81
Period 3 European
call
Period 3 European
call
Problem 10.20.
www.actuary88.com
c
Yufeng
Guo
82
e(r)h d
e(0.050.03)1 0.755 784
=
= 0.425 557
ud
1. 377 128 0.755 784
Vdd =?
Sddd = 100 0.755 7843 = 43. 171 1
Vddd = 0
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.05)1 (0.425 557 166. 169 8 + 0.574 443 48. 333) =
93. 676 4
R
Period 1
2
Vuu = 94. 648 2
c
Yufeng
Guo
83
1
Vu = 49. 004 8
S = 100
EV = 100 95 = 5
V =?
Vd = 7. 920 1
V R = erh (u Vu + d Vu ) = e(0.05)1 (0.425 557 49. 004 8 + 0.574 443 7. 920 1) =
24. 165 0
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.05)1 (0.425 557 0 + 0.574 443 0) =
R
Vud = max Vud
, EVud = 8. 927 0
R
Vdd
= erh ( u Vudd + d Vddd ) = e(0.05)1 (0.425 557 16. 337 1 + 0.574 443 51. 828 9) =
34. 934 0
R
www.actuary88.com
c
Yufeng
Guo
84
2
Vuu = 0
1
Vu = 4. 8780
S = 100
EV = 0
V =?
Vd = 24. 311 8
V R = erh ( u Vu + d Vu ) = e(0.05)1 (0.425 557 4. 8780 + 0.574 443 24. 311 8) =
15. 259 3
c. If we switch S and K and switch r and and recalculate the option price,
what happens?
After the switch, we have:
S = 95
K = 100
r = 3%
= 5%
c
Yufeng
Guo
85
u =
e(r)h d
e(0.030.05)1 0.726 149
=
= 0.425 557
ud
1. 323 13 0.726 149
Vdd =?
Sddd = 95 0.726 1493 = 36. 374 8
Vddd = 0
R
Vuu
= erh ( u Vuuu + d Vuud ) = e(0.03)1 (0.425 557 120. 055 + 0.574 443 20. 768 7) =
61. 158 1
R
Period 1
2
Vuu = 66. 313 9
c
Yufeng
Guo
86
1
Vu = 32. 167 7
S = 95
EV = 0
V =?
Vd = 3. 542 2
V R = erh ( u Vu + d Vu ) = e(0.03)1 (0.425 557 32. 167 7 + 0.574 443 3. 542 2) =
15. 2593
R
Vuu
= erh (u Vuuu + d Vuud ) = e(0.03)1 (0.425 557 0 + 0.574 443 0) =
R
Vuu = max Vuu
, EVuu = 0
R
Vud
= erh ( u Vuud + d Vudd ) = e(0.03)1 (0.425 557 0 + 0.574 443 33. 720 7) =
18. 798 1
R
c
Yufeng
Guo
87
2
Vuu = 0
1
Vu = 10. 479 3
S = 95
EV = 100 95 = 5
V =?
Vd = 35. 584 8
V R = erh (u Vu + d Vu ) = e(0.03)1 (0.425 557 10. 479 3 + 0.574 443 35. 584 8) =
24. 165 04
So the American put premium is $24. 165 0, which is equal to the American
call premium before the switch.
Problem 10.21.
Suppose u < e(r)h . Since d < u, we have d < u < e(r)h . This means
that the savings account is always better than the stock. So at t = 0, we short
sell eh share of stock and investment the short sale proceeds Seh into the
savings account. Then at time h, we close our short position by buying one
stock from the market. The stock price at time h is either uS or dS.
t=0
t = h, u mode
t = h, d mode
short eh stock
Seh
uS
dS
deposit Seh in savings Seh Se(r)h
Seh
h
Total
0
S e
u > 0 S eh d > 0
c
Yufeng
Guo
88
h
Total
0
S ue
> 0 S d eh > 0
So initial cost is zero yet we have positive payo at time h. This is an
arbitrage opportunity.
www.actuary88.com
c
Yufeng
Guo
89
www.actuary88.com
c
Yufeng
Guo
90
Chapter 11
S = 100
EV = 100 70 = 30
V
t=1
Su = 100 (1. 246 077) = 124. 607 7
Vu = 124. 607 7 70 = 54. 607 7
Sd = 100 (0.683 861 ) = 68. 386 1
Vd = 0
S = 100
EV = 100 80 = 20
V
t=1
Su = 100 (1. 246 077) = 124. 607 7
Vu = 124. 607 7 80 = 44. 607 7
Sd = 100 (0.683 861 ) = 68. 386 1
Vd = 0
91
S = 100
EV = 100 90 = 10
V
t=1
Su = 100 (1. 246 077) = 124. 607 7
Vu = 124. 607 7 90 = 34. 607 7
Sd = 100 (0.683 861 ) = 68. 386 1
Vd = 0
S = 100
EV = 100 100 = 0
V
t=1
Su = 100 (1. 246 077) = 124. 607 7
Vu = 124. 607 7 100 = 24. 607 7
Sd = 100 (0.683 861 ) = 68. 386 1
Vd = 0
c
Yufeng
Guo
92
c
Yufeng
Guo
93
S = 100
EV = 100 70 = 30
V
t=1
Su = 100 (1. 349 859) = 134. 985 9
Vu = 134. 985 9 70 = 64. 985 9
Sd = 100 (0.740 818 ) = 74. 081 8
Vd = 74. 081 8 70 = 4. 081 8
S = 100
EV = 100 80 = 20
V
t=1
Su = 100 (1. 349 859) = 134. 985 9
Vu = 134. 985 9 80 = 54. 985 9
Sd = 100 (0.740 818 ) = 74. 081 8
Vd = 0
S = 100
EV = 100 90 = 10
V
t=1
Su = 100 (1. 349 859) = 134. 985 9
Vu = 134. 985 9 90 = 44. 985 9
Sd = 100 (0.740 818 ) = 74. 081 8
Vd = 0
c
Yufeng
Guo
94
S = 100
EV = 0
V
t=1
Su = 100 (1. 349 859) = 134. 985 9
Vu = 134. 985 9 100 = 34. 985 9
Sd = 100 (0.740 818 ) = 74. 081 8
Vd = 0
0.08
PEur < 100 100e
K 1 e0.08 = 7. 69 0.0769K
Using the BlackScholes option pricing formula, we find the following put
price (S = 100, T = 1, r = 0.08, = 0, = 0.3):
K
70
80
90
100
PEur
0.7752
2.0904
4.4524
8.0229
7. 69 0.0769K
7. 69 0.0769 (70) = 2. 307
7. 69 0.0769 (80) = 1. 538
7. 69 0.0769 (90) = 0.769
7. 69 0.0769 (100) = 0
Among the 4 strike prices, only K = 70 satisfies the condition PEur < 7.
69 0.0769K. Hence of the 4 strike prices given, only K = 70 leads to optimal
early exercise.
Problem 11.3.
If = 0, then the stock doesnt pay any dividend. Its never optimal to early
exercise an American call on a nondividend paying stock. So early exercise will
never occur.
www.actuary88.com
c
Yufeng
Guo
95
S = 100
EV = 0
V
t=1
Su = 100 (1. 462 285) = 146. 228 5
Vu = 0
S = 100
EV = 110 100 = 10
V
t=1
Su = 100 (1. 462 285) = 146. 228 5
Vu = 0
S = 100
EV = 120 100 = 20
V
www.actuary88.com
t=1
Su = 100 (1. 462 285) = 146. 228 5
Vu = 0
96
t=1
Su = 100 (1. 462 285) = 146. 228 5
Vu = 0
S = 100
EV = 130 100 = 30
V
Using the BlackScholes option pricing formula, we find the following put
price (S = 100, T = 1, r = 0.08, = 0, = 0.3):
K
100
110
120
130
CEur
15.7113
11.2596
7.8966
5.4394
0.07 69K
0.07 69 (100) = 7. 69
0.07 69 (110) = 8. 459
0.07 69 (120) = 9. 228
0.07 69 (130) = 9. 997
Among the 4 strike prices, only K = 130 satisfies the condition CEur < 0.07
69K. Hence of the 4 strike prices given, only K = 130 leads to optimal early
exercise.
www.actuary88.com
c
Yufeng
Guo
97
c
Yufeng
Guo
98
c
Yufeng
Guo
99
We are reusing the data in the textbook Figure 11.4 except that we set
S = 100. We have:
S = 100
r = 0.08
= 0.3
=0
T =1
h = T /n = 1/3
Then:
c
Yufeng
Guo
100
3
= 0.456
141 7)+3 0.456 8062 (0.543 194) 128. 814 8+3 (0.456 806) 0.543 1942 91.
806 (182.
c
Yufeng
Guo
101
0.25
0.5
0.75
1
95.71879
79.73044
66.41269
55.31947
46.07921
69.12305
57.57710
47.95973
39.94880
49.91701
41.57914
34.63398
36.04742
30.02625
26.03154
www.actuary88.com
c
Yufeng
Guo
102
d
P
P
Ft+h,T
= Ft,T
d = 46. 079 21 0.866 959 = 39. 948 8
The prepaid
ud
P
P
Ft+2h,T
= Ft,T
ud = 46. 079 21 1. 200 53 0.866 959 = 47. 959 73
dd
P
P 2
Ft+2h,T
= Ft,T
d = 46. 079 21 0.866 9592 = 34. 633 98
(My numbers are calculated using Excel so you might not be able to fully
match mine.)
Next, convert the prepaid forward price tree into a stock price tree. The
onetoone mapping between the prepaid forward
price and the stock price is
Der(TD tt) if TD t + t
P
P
St+t = Ft+t,T +P V (Div) = Ft+t,T +
0
if TD < t + t
The PV of the dividend at each interval is:
Time
0
0.25 0.5 0.75
Dividend time
0.25
Dividend amount
$4
P V (Div)
3. 920 79 4
0
0
1
95.71879
79.73044
66.41269
55.31947 + 4
46.07921 + 3. 920 79
69.12305
57.57710
47.95973
39.94880 + 4
49.91701
41.57914
34.63398
36.04742
30.02625
26.03154
c
Yufeng
Guo
103
0.25
0.5
0.75
1
95.718794
79.730440
66.412694
59.319474
50
69.123048
57.577104
47.959733
43.948797
49.917007
41.579138
34.633981
36.047421
30.026253
26.031540
c
Yufeng
Guo
104
0.25
0.5
0.75
1
50.71879
35.62150
23.17717
max (14.27212, 14.31947)
8.45513
24.12305
13.46816
7.23801
3.78761
4.91701
2.21414
0.99703
0
0
0
Its optimal to exercise the American call at the upper node at t = 0.25. The
roll back value is 14.27212. The exercise value is 59.319474 45 = 14. 319 474,
which is greater than the roll back value. The premium of 8.45513 is calculated
as follows:
The roll back value is: e0.08(0.25) (14. 319 474 0.459 399 + 3.78761 0.540 601) =
8. 455 132 8
The early exercise value at t = 0 is 50 45 = 5.
We take the greater of the two. So the American call premium is 8.45513.
By the way, if you bother to calculate the European put and the American
put premium with strike price K = 45, here are the results:
The prepaid forward price tree and the stock price tree wont change whether
the option is a call or put.
The European put premium:
Time 0 0.25
0.5
0.75
1
0
0
0
1.33205
3.89484
0
2.51380
6.21822
0
4.74394
9.59857
8.95258
14.08269
18.96846
c
Yufeng
Guo
105
0.25
0.5
0.75
1
0
0
0
1.33205
4.11033
0
2.51380
6.62489
0
4.74394
8.95258
10.36602
14.97375
18.96846
The two bold numbers indicate that early exercise is optimal.
The American put premium is 4.11033.
www.actuary88.com
c
Yufeng
Guo
106
Chapter 12
BlackScholes formula
Problem 12.1.
Skip this spreadsheet problem.
Problem 12.2.
Skip this problem but remember the following key point. As n gets bigger,
the price calculated using the discrete binomial tree method approach the price
calculated using the BlackScholes formula.
Problem 12.3.
a. r = 8%
=0
T
European call price
1
7.8966
10
56.2377
100
99.9631
1, 000
100
1, 0000 100
1, 0000 100
As T , the European call premium approaches the current stock price.
b. r = 8%
T
1
10
100
1, 000
1, 0000
10, 0000
= 0.1%
European call price
7.8542
55.3733
90.4471
36.7879
0.0045
0
107
c
Yufeng
Guo
108
95
1
1
S
ln
+ r + 2 T
+ 0.015 0.035 + 0.12 0.5
K
2
90
2
=
=
0.1
0.5
T
1/95
S
1 2
1
2
ln
ln
+ r+ T
+ 0.035 0.015 + 0.1 0.5
K
2
1/90
2
=
d1 =
=
0.1 0.5
T
0.587 849
N (d1 ) = 0.278 316 8
1 0.015(0.5)
1
0.278 316 8 e0.035(0.5)
e
95
90
c
Yufeng
Guo
109
1
euro and get 1 yen" at T = 0.5 is worth 0.0001
90
226 1 (euro) at t = 0 (statement b)
1
This can be expressed at
1Y = 0.0001 226 1
90
The privilege of "give
90
90
90
The above equation means this: If you can sell 1 for a guaranteed price
1
for a guaranteed price of 1Y . This
90Y , then you must be able to sell
90
should make intuitive sense.
1
1. 048 3
Y into euros. Since at t = 0, 1=95Y or 1Y =
,
We convert
90
95
then
1. 048 3
1
1. 048 3
Y =
= 0.0001226 1
90
90 95
1
1. 048 3
1
1Y =
(1 90Y ) =
Y = 0.0001226 1
90
90
90
This is exactly Statement b
Next,
we derive
Statement b from Statement a.
1
1Y = 0.0001 226 1 (statement b)
90
1
1Y = 90 (0.0001 226 1) = 90 (0.0001 226 1) 95Y =
(1 90Y ) = 90
90
1. 048 3Y
This is exactly Statement a.
Problem 12.6.
www.actuary88.com
c
Yufeng
Guo
110
100
1
1
S
ln
+ r + 2 T
+ 0.06 0 + 0.42 1
ln
K
2
105
2
=
= 0.228 024 589
d1 =
0.4 1
T
0.228 025
N (d1 ) = 0.590 186 6
1
S
F
100e(0.060)1 1
1 2
+ r + 2 T
ln
ln
T
+
+ 0.42 (1)
K
2
K 2
105
2
d1 =
=
=
=
0.4 1
T
T
1
100
+ 0.06 0 + 0.42 1
ln
105
2
S
1
1
Se(r)T
ln
+ r r + 2 T
+ r + 2 T
ln
K
2
K
2
d1 =
=
T
T
d2 = d1 T
www.actuary88.com
c
Yufeng
Guo
111
same call
P = Se(r)T erT N (d1 ) + KerT N (d2 ) = SeT N (d1 ) +
KerT N (d2 )
This is the same put formula when the underlying asset is a stock.
We can also prove that the futures option premium is equal to the underlying
stock option premium intuitively. On the maturity date T , the futures price is
equal to the stock price. So if we stand at T , the payo of a futures option and
the payo of an otherwise identical stock option are identical. Consequently, the
premium of a futures option is equal to the premium of an otherwise identical
stock option.
Problem 12.7.
a.
100
1
1
S
ln
+ r + 2 T
+ 0.08 0.03 + 0.32 0.75
K
2
95
2
d1 =
=
=
0.3 0.75
T
0.471 669
b.
1
100e0.03(0.75)
1
S
2
+
0
0
+
0.75
ln
+ r + 2 T
0.3
K
2
2
95e0.08(0.75)
=
d1 =
=
0.3 0.75
T
0.471 669100
1 2
SeT
S
1
+
0
0
+
ln
T
ln
+ r + 2 T
K
2
KerT
2
=
d1 =
T
T
d2 = d1 T
www.actuary88.com
c
Yufeng
Guo
112
103. 821 20
S
1
1
ln
ln
+ r + 2 T
+ 0.08 0.08 + 0.32 0.75
K
2
95
2
=
=
d1 =
0.3 0.75
T
0.471 669
d1 =
=
= 1. 154 1
0.3
0.5
ln
N (d1 ) = 0.875 77
N (d2 ) = 0.826 90
C = 48 (0.875 77) 40e0.08(0.5) (0.826 90) = 10. 258
So the European call premium is 10. 258
www.actuary88.com
c
Yufeng
Guo
113
d1 =
=
= 1. 154 1
0.3 0.5
T
ln
N (d1 ) = 0.875 77
N (d2 ) = 0.826 90
C = 48 (0.875 77) 40e0.08(0.5) (0.826 90) = 10. 258
The exercise value is EV = 60 40 = 20 > C
Hence its optimal to early exercise the American put at t = 0.
c. Its optimal to early exercise the American call at t = 0 if the exercise
value is greater than the European call premium. Its not optimal to exercise
the American call at t = 0 if the exercise value is equal to or less than the he
European call premium. However, keep in mind that if the stock doesnt pay
dividend, then its never optimal to early exercise an American call.
Problem 12.10.
The statement means that the absolute value  (t)  =  V
t  reaches its maximum value when t T . In other words, the closer to the expiration date, the
higher the  (t) . This statement is not correct. I dont know of any intuitive
way to explain why this statement is not correct. Thats why the textbook asks
you to test this statement using a spreadsheet.
www.actuary88.com
c
Yufeng
Guo
114
( )
The formula V ega = V (+ )V
is an approximation. For this approx2
imation to work, needs to be small. Since Appendix 12.A is excluded from
both CAS and SOA exam, you can ignore this part.
b. If you want to solve this problem, you can set up some test cases and
compare the approximated Vega with the actual Vega (using the spreadsheet).
For the purpose of passing the exam, you can ignore this part.
Problem 12.12.
Since Appendix 12.A is excluded from both CAS and SOA exam, you can
ignore this problem.
Problem 12.13.
Lets not worry about drawing a diagram and focus on how to calculate the
profit. Ill do some sample calculations.
Lets calculate the profit after 6 months (i.e. at expiration) assuming the
stock price after 6 months is $60.
Using the BlackScholes formula, we can find:
The 40strike call premium is 4.1553. This call premium is calculated
using the BlackScholes formula by setting S = 40, K = 40, T = 0.5, r =
0.08, = 0.3, = 0
The 45strike call premium is 2.1304. This call premium is calculated
using the BlackScholes formula by setting S = 40, K = 45, T = 0.5, r =
0.08, = 0.3, = 0
The net cost of the bull spread at t = 0 is 4.1553 2.1304 = 2. 024 9
Its future value at expiration is 2. 024 9e0.08(0.5) = 2. 107 5
The stock price at expiration is 60.
www.actuary88.com
c
Yufeng
Guo
115
c
Yufeng
Guo
116
S = 40
1
Price
Delta
Gamma
Vega
Theta
Rho
Pn
1
1
1
1
1
1
i=1
i Greeki
Greek1
bought 40K call
4.1553
0.6159
0.0450
0.1080
0.0134
0.1024
2
1
1
1
1
1
1
Greek2
sold 45K call
2.1304
0.3972
0.0454
0.1091
0.0120
0.0688
Pn
Greekoption = i=1 i Greeki
Portfolio Price and Greeks
1 (4.1553) + 1 (2.1304) = 2. 024 9
1 (0.6159) + 1 (0.3972) = 0.218 7
1 (0.0450) + 1 (0.0454) = 0.000 4
1 (0.1080) + 1 (0.1091) = 0.001 1
1 (0.0134) + 1 (0.0120) = 0.001 4
1 (0.1024) + 1 (0.0688) = 0.033 6
c
Yufeng
Guo
117
S = 45
1
Price
Delta
Gamma
Vega
Theta
Rho
1
1
1
1
1
1
Greek1
bought 40K call
7.7342
0.8023
0.0291
0.0885
0.0135
0.1418
2
1
1
1
1
1
1
Greek2
sold 45K call
4.6747
0.6159
0.0400
0.1216
0.0150
0.1152
P
Greekoption = ni=1 i Greeki
Portfolio Price and Greeks
1 (7.7342) + 1 (4.6747) = 3. 059 5
1 (0.8023) + 1 (0.6159) = 0.186 4
1 (0.0291) + 1 (0.0400) = 0.010 9
1 (0.0885) + 1 (0.1216) = 0.033 1
1 (0.0135) + 1 (0.0150) = 0.001 5
1 (0.1418) + 1 (0.1152) = 0.026 6
c. Ignore this part. Not sure what this problem wants to accomplish.
Problem 12.15.
a.
S = 40
1
Price
Delta
Gamma
Vega
Theta
Rho
b.
1
1
1
1
1
1
S = 45
1
Price
Delta
Gamma
Vega
Theta
Rho
1
1
1
1
1
1
Greek1
bought 40K call
2.5868
0.3841
0.0450
0.1080
0.0049
0.0898
Greek1
bought 40K call
1.1658
0.1977
0.0291
0.0885
0.0051
0.0503
1
1
1
1
1
1
1
1
1
1
1
1
Greek2
sold 45K call
5.3659
0.6028
0.0454
0.1091
0.0025
0.1474
P
Greekoption = ni=1 i Greeki
Portfolio Price and Greeks
1 (2.5868) + 1 (5.3659) = 2. 779 1
1 (0.3841) + 1 (0.6028) = 0.218 7
1 (0.0450) + 1 (0.0454) = 0.000 4
1 (0.1080) + 1 (0.1091) = 0.001 1
1 (0.0049) + 1 (0.0025) = 0.002 4
1 (0.0898) + 1 (0.1474) = 0.057 6
Greek2
sold 45K call
2.9102
0.3841
0.0400
0.1216
0.0056
0.1010
Pn
Greekoption = i=1 i Greeki
Portfolio Price and Greeks
1 (1.1658) + 1 (2.9102) = 1. 744 4
1 (0.1977) + 1 (0.3841) = 0.186 4
1 (0.0291) + 1 (0.0400) = 0.010 9
1 (0.0885) + 1 (0.1216) = 0.033 1
1 (0.0051) + 1 (0.0056) = 0.000 5
1 (0.0503) + 1 (0.1010) = 0.050 7
c. Ignore
Problem 12.16.
Theres no easy way to solve this problem manually. This type of problems
shouldnt show up in the exam.
Problem 12.17.
Theres no easy way to solve this problem manually. This type of problems
shouldnt show up in the exam.
www.actuary88.com
c
Yufeng
Guo
118
K = 60
r = 0.06
= 0.4
= 0.03
a.
Solve for h.
1 2
h (h 1) + (r ) h r = 0
2
1
0.42 h (h 1) + (0.06 0.03) h 0.06 = 0
2
h = 0.608 182 5
h = 1. 233 182 5
Use the bigger h for call and the smaller h for put.
hcall
1. 233 182 5
HCall
=
K=
60 = 317. 309 19
hcall 1
1. 233 182 5 1
K)
Cperpetual = (HCall
26. 351 83
HCall
hcall
50
317. 309 19
1. 233 182 5
The call should be exercised when the stock price reaches 317. 309 19. This
price is called the barrier.
The call premium is 26. 351 83.
b. Now = 0.04 (instead of = 0.03). Everything else is the same as in a.
Higher dividend yield means that the stock price will decrease more quickly.
Recall that the stock price drops by the dividend amount immediately after
the dividend payment time. We expect that under = 0.04 the optimal stock
price (i.e. barrier) is lower than the barrier when = 0.03. In addition, we
expect that the option price under = 0.04 is lower than the option price under
= 0.03.
Solve for h.
1 2
h (h 1) + (r ) h r = 0
2
1
0.42 h (h 1) + (0.06 0.04) h 0.06 = 0
2
h = 0.568 729 3
h = 1. 318 729 3
Use the bigger h for call and the smaller h for put.
hcall
1. 318 729 3
HCall
=
K=
60 = 248. 247 52
hcall 1
1. 318 729 3 1
Cperpetual =
(HCall
22. 751 3
K)
HCall
hcall
50
248. 247 52
1. 318 729 3
The call should be exercised when the stock price reaches 248. 247 52.
The call premium is 22. 751 28
c.Now r = 0.07 (instead of r = 0.06). Everything else is the same as in a.
www.actuary88.com
c
Yufeng
Guo
119
=
HCall
K=
60 = 334. 919 4
hcall 1
1. 218 245 8 1
Cperpetual =
(HCall
K)
27. 100 1
HCall
hcall
50
334. 919 4
1. 218 245 8
The call should be exercised when the stock price reaches 334. 919 4
The call premium is 27. 1001
d.
The higher the volatility, the more valuable an option is. As goes
up from 0.4 to 0.5, we expect both the barrier and the call option premium will
go up.
1 2
h (h 1) + (r ) h r = 0
2
1
0.52 h (h 1) + (0.06 0.03) h 0.06 = 0
2
h = 1. 170 189 9
h = 0.410 189 9
Use the bigger h for call and the smaller h for put.
hcall
1. 170 189 9
=
K=
60 = 412. 547 36
HCall
hcall 1
1. 170 189 9 1
K)
Cperpetual = (HCall
29. 835 5
HCall
hcall
50
412. 547 36
1. 170 189 9
The call should be exercised when the stock price reaches 412. 547 36
The call premium is 29. 835 5
Problem 12.19.
S = 50
K = 60
a.Solve for h.
www.actuary88.com
r = 0.06
= 0.4
= 0.03
1 2
h (h 1) + (r ) h r = 0
2
c
Yufeng
Guo
120
S
HP ut
hput
50
22. 690 80
0.608 182 5
The put should be exercised when the stock price reaches 22. 690 80.
The put premium is 23. 074 7
b. Now = 0.04 (instead of = 0.03). Everything else is the same as in a.
Higher dividend yield means that the stock price will decrease more quickly.
Since the value of a put goes up if the stock price goes down, we expect that the
barrier price will go down (i.e. exercise occurs later because we want to exercise
when the stock price is low) and the put premium will go up.
1 2
Solve for h.
h (h 1) + (r ) h r = 0
2
1
0.42 h (h 1) + (0.06 0.04) h 0.06 = 0
2
h = 0.568 729 3
h = 1. 318 729 3
Use the bigger h for call and the smaller h for put.
hput
0.568 729 3
K=
60 = 21. 752 5
HP ut =
hcall 1
0.568 729 3 1
Pperpetual = (K HP ut )
23. 824 8
S
HP ut
hput
50
21. 752 5
0.568 729 3
The put should be exercised when the stock price reaches 21. 752 5
The put premium is 23. 824 8
c. As r goes up, people expect to get increased return from the stock. The
stock price goes up and the put option premium goes down.
We expect the barrier will go up (compared with a, meaning that exercise
occurs sooner).
1 2
h (h 1) + (r ) h r = 0
2
1
0.42 h (h 1) + (0.07 0.03) h 0.07 = 0
2
h = 1. 218 245 8
h = 0.718 245 8
Use the bigger h for call and the smaller h for put.
hput
0.718 245 8
K=
60 = 25. 080 67
HP ut =
hcall 1
0.718 245 8 1
www.actuary88.com
c
Yufeng
Guo
121
Pperpetual = (K
HP ut )
23. 824 8
S
HP ut
hput
50
25. 080 67
0.568 729 3
The put should be exercised when the stock price reaches 25. 080 67
The put premium is 23. 824 8
d.
The higher the volatility, the more valuable an option is. As goes
up from 0.4 to 0.5, we expect both the barrier will go down (i.e. exercise occurs
later) and the put option premium will go up.
1 2
h (h 1) + (r ) h r = 0
2
1
0.52 h (h 1) + (0.06 0.03) h 0.06 = 0
2
h = 1. 170 189 9
h = 0.410 189 9
Use the bigger h for call and the smaller h for put.
hput
0.410 189 9
HP ut =
K=
60 = 17. 452 5
hcall 1
0.410 189 9 1
Pperpetual = (K HP ut )
27. 629 4
S
HP ut
hput
50
17. 452 5
0.410 189 9
The put should be exercised when the stock price reaches 17. 452 5
The put premium is 27. 629 4
Problem 12.20.
For a and b, if you use the BlackScholes formula, youll find that call and
put are both worth 17.6988.
For part c. After we switch S and K and switch r and , the put after
the switch and the call before the switch have the same value. This is not a
coincidence. Its explained in my solution to Problem 10.19.
Problem 12.21.
1 2
a.
h (h 1) + (r ) h r = 0
2
1
0.32 h (h 1) + (0.08 0.05) h 0.08 = 0
2
h = 1. 177 043
h = 1. 510 376
Use the bigger h for call and the smaller h for put.
hcall
1. 510 376
=
HCall
K=
90 = 266. 340 6
hcall 1
1. 510 376 1
www.actuary88.com
c
Yufeng
Guo
122
Cperpetual =
(HCall
K)
HCall
40. 158 9
The call premium is 40. 158 9
hcall
100
266. 340 6
1. 510 376
1 2
b.
h (h 1) + (r ) h r = 0
2
1
0.32 h (h 1) + (0.05 0.08) h 0.05 = 0
2
h = 0.510 38
h = 2. 177 04
Use the bigger h for call and the smaller h for put.
hput
0.510 38
HP ut =
K=
100 = 33. 791 5
hcall 1
0.510 38 1
Pperpetual = (K
HP ut )
St
HP ut
40. 158 9
The put premium is 40. 158 9
hput
90
33. 791 5
0.510 38
www.actuary88.com
c
Yufeng
Guo
123
www.actuary88.com
c
Yufeng
Guo
124
Chapter 13
91
40
S
1
1
ln
+ r + 2 T
+ 0.08 0 + 0.32
K
2
45
2
365
r
=
=
d1 =
T
91
0.3
365
0.578 25
r
91
= 0.728 04
d2 = d1 T = 0.578 25 0.3
365
N (d1 ) = 0.281 55
N (d2 ) = 0.233 29
C = SeT N (d1 ) KerT N (d2 ) = 40e0(91/365) 0.281 55 45e0.08(91/365)
0.233 29 = 0.971 3
ln
as follows:
90
39
1 2
1
S
2
ln
+ r+ T
+ 0.08 0 + 0.3
ln
K
2
45
2
365
r
d1 =
=
=
T
90
0.3
365
0.753 706
r
90
d2 = d1 T = 0.753 706 346 0.3
= 0.902 675
365
N (d1 ) = 0.774 49
N (d2 ) = 0.183 349
C = SeT N (d1 ) KerT N (d2 ) = 39e0(90/365) 0.225 51345e0.08(90/365)
0.183 349 = 0.705 5
A call on 100 stocks is worth 100 (0.705 5) = 70. 55
So the trader needs to pay 70. 55 to buy a new call to cancel out the original
call he sold.
At t = 0, the trader owns 28. 155 stocks. Now one day later, the stocks
are worth 28. 155 (39) = 1098. 045. So the trader sells out his stocks, receiving
1098. 05.
In addition, the trader needs to pay back the loan borrowed from the bank.
The future value of the loan one day later is:
1029. 07e0.08(1/365) = 1029. 30
So the traders net wealth at t = 1/365 is:
1098. 05 (1029. 30 + 70. 55) = 1. 8
So the trader lost $1.8.
If you want to following the textbook calculation on Page 417 "Day 1:
Markingtomarket," here it is:
Day 1
Gain on 28. 155 shares
28. 155 (39 40) = 28. 155
Gain on the written call 97. 13 70. 55 = 26. 58
Interest
1029. 07 e0.08(1/365) 1 = 0.226
Overnight profit
28. 155 + 26. 58 0.226 = 1. 8
If the stock price is 40.5 on Day 1.
To close his short call position, the trader can buy, from the open market,
a 45strike call expiring in 90 days. This new (purchased) call and the original
(sold) call have the same underlying, same expiration date, same strike price.
www.actuary88.com
c
Yufeng
Guo
126
as follows:
40.5
90
S
1 2
1
2
ln
ln
+ r+ T
+ 0.08 0 + 0.3
K
2
45
2
365
r
=
=
d1 =
90
T
0.3
365
0.500 363
r
90
d2 = d1 T = 0.500 363 0.3
= 0.649 332
365
N (d1 ) = 0.308 41
N (d2 ) = 0.258 062
C = SeT N (d1 )KerT N (d2 ) = 40.5e0(90/365) 0.308 4145e0.08(90/365)
0.258 062 = 1. 104 65
A call on 100 stocks is worth 100 (1. 104 65) = 110. 465
One Day 1, the trader
pays 110. 46 and buy a new call to cancel out the original call he sold
sells out his 28. 155 stocks, receiving 28. 155 (40.5) = 1140. 28
pays 1029. 07e0.08(1/365) = 1029. 30 to the bank to payo the loan
So the traders net wealth on Day 1 is:
1140. 28 (1029. 30 + 110. 46) = 0.52
If you want to use the textbook notation, here you go:
Day 1
Gain on 28. 155 shares
28. 155 (40.5 40) = 14. 077 5
Gain on the written call 97. 13 110. 465 = 13.335
Interest
1029. 07 e0.08(1/365) 1 = 0.226
Overnight profit
14. 077 5 13. 335 0.226 = 0.52
So the trader gained $0.52.
Problem 13.2.
91
40
1
1
S
ln
+ r + 2 T
+ 0.08 0 + 0.32
K
2
40
2
365
r
=
=
d1 =
T
91
0.3
365
0.208 05
r
91
d2 = d1 T = 0.208 05 0.3
= 0.05 825
365
N (d2 ) = 0.476 77
N (d1 ) = 0.417 59
P = 40e0.08(91/365) 0.476 77 40e0(91/365) 0.417 59 = 1. 990 6
ln
www.actuary88.com
c
Yufeng
Guo
127
new put is
90
39
S
1 2
1
ln
ln
+ r+ T
+ 0.08 0 + 0.32
K
2
40
2
365
r
d1 =
= 0.0
=
T
90
0.3
365
3695
r
90
d2 = d1 T = 0.03695 0.3
= 0.112 02
365
N (d2 ) = 0.544 60
N (d1 ) = 0.485 26
P = KerT N (d2 )SeT N (d1 ) = 40e0.08(90/365) 0.544 6039e0(90/365)
0.485 26 = 2. 433 4
On Day 1 a put on 100 stocks is worth 100 (2. 433 4) = 243. 34
One Day 1, the trader
pays 243. 34, buying a new put to cancel out the original put he sold
buys back 41. 76 stocks to close the short sale position, paying 41. 76 (39) =
1628. 64
receives 1869. 46e0.08(1/365) = 1869. 87 from the bank
www.actuary88.com
c
Yufeng
Guo
128
90
40.5
S
1 2
1
2
ln
ln
+ r+ T
+ 0.08 0 + 0.3
K
2
40
2
365
r
d1 =
=
=
T
90
0.3
365
0.290 29
r
90
d2 = d1 T = 0.290 29 0.3
= 0.141 32
365
N (d1 ) = 0.385 80
N (d2 ) = 0.443 81
P = KerT N (d2 )SeT N (d1 ) = 40e0.08(90/365) 0.443 8140.5e0(90/365)
0.385 80 = 1. 780 8
On Day 1 a put on 100 stocks is worth 100 (1. 780 8) = 178. 08
One Day 1, the trader
pays 178. 08, buying a new put to cancel out the original put he sold
buys back 41. 76 stocks to close the short sale position, paying 41. 76 (40.5) =
1691. 28
receives 1869. 46e0.08(1/365) = 1869. 87 from the bank
So the traders net wealth on Day 1 is:
178. 08 1691. 28 + 1869. 87 = 0.51
So the trader gains 0.51.
If you want to use the textbook notation, here you go:
Day 1
Gain on 41. 76 shares
41. 76 (40 40.5) = 20. 88
Gain on the written put 199. 06 178. 08 = 20. 98
Interest income
1869. 46 e0.08(1/365) 1 = 0.41
Overnight profit
20. 88 + 20. 98 + 0.41 = 0.51
www.actuary88.com
c
Yufeng
Guo
129
13.2 theres
91
40
S
1 2
1
ln
ln
+ r+ T
+ 0.08 0 + 0.32
K
2
45
2
365
r
=
=
d1 =
T
91
0.3
365
0.578 25
r
91
d2 = d1 T = 0.578 25 0.3
= 0.728 04
365
N (d1 ) = 0.718 45
N (d2 ) = 0.766 71
P = KerT N (d2 )SeT N (d1 ) = 45e0.08(91/365) 0.766 7140e0(91/365)
0.718 45 = 5. 082 6
A put on 100 stocks is worth 100 (5. 082 6) = 508. 26
Use the formula in Appendix 12.8 to find
= SeT N (d1 ) = e091/365 0.718 45 = 0.718 45 (negative delta
means short selling stocks)
Suppose a trader sells a put option on 100 stocks. To hedge his risk, the
trader should at t = 0
sell the call and receive 100 (5. 082 6) = 508. 26
buy 0.718 45 (100) = 71. 845 stocks (i.e. short sell 71. 845 stocks) receiving 71. 845 (40) = 2873. 8
lend 2873. 8 + 508. 26 = 3382. 06 to a bank
The traders net position is zero at t = 0
If the stock price is 39.5 on Day 1.
To close his short put position, the trader can buy, from the open market,
a 45strike put expiring in 90 days. This new (purchased) put and the original
(sold) put have the same underlying, same expiration date, same strike price.
They will cancel out each other and the trader doesnt have any liabilities associated either put.
The cost of thenew call is 110.46.
Its calculated
as follows:
90
39
S
1 2
1
2
ln
ln
+ r+ T
+ 0.08 0 + 0.3
K
2
45
2
365
r
d1 =
=
=
T
90
0.3
365
0.753 706
r
90
d2 = d1 T = 0.753 706 346 0.3
= 0.902 675
365
N (d1 ) = 0.774 49
www.actuary88.com
c
Yufeng
Guo
130
as follows:
90
40.5
1 2
1
S
2
ln
+ r+ T
+ 0.08 0 + 0.3
ln
K
2
45
2
365
r
d1 =
=
=
T
90
0.3
365
0.500 363
r
90
= 0.649 332
d2 = d1 T = 0.500 363 0.3
365
N (d1 ) = 0.691 59
N (d2 ) = 0.741 94
P = KerT N (d2 )SeT N (d1 ) = 45e0.08(90/365) 0.741 9440.5e0(90/365)
0.691 59 = 4. 725 76
On Day 1 a put on 100 stocks is worth 100 (4. 725 76) = 472. 576
One Day 1, the trader
www.actuary88.com
c
Yufeng
Guo
131
Delta
0.58240
0.28155
0.28155 + 0.58240 = 0.300 85
100 (0.300 85) = 30. 085
c
Yufeng
Guo
132
c
Yufeng
Guo
133
Delta
(0.71845) = 0.718 451
0.41760
0.71845 0.41760 2 = 0.116 75
100 (0.116 75) = 11. 675
www.actuary88.com
c
Yufeng
Guo
134
=0
c
Yufeng
Guo
135
interest earned
Interest earned at the end of Day 0: M V (0) erh 1 = (1869. 434) e0.08/365 1 =
0.409 78 = 0.41
The investment at the beginning of Day 0 is: M V (0) = 1869. 434
Day 1:
In the beginning of Day 1, the trader starts from a clean slate. He buys sells
1 = 100 (0.385797) = 38. 579 7 stocks (i.e. short sell stocks) and receives
the put premium C1 = $178.08.
His asset (or investment) is M V (1) = 1 S1 C1 = (38. 579 7) 40.5
178.08 = 1740. 557 85
The traders asset at the end of Day 1 before he rebalances the portfolio (i.e.
before he starts over from a clean slate the next day) is:
M V BR (2) = 1 S2 C2 = (38. 579 7) 39.25 230.55 = 1744. 803 225
The traders profit at the end of Day 1 is:
M V BR (2) M V (1) erh = 1744. 803 225 (1740. 557 85) e0.08(1/365) =
3. 864
To find the capital gain and the interest earned at the end of Day 1, we just
need to break down the profit at the end of Day 1 into two parts:
interest earned
Ill omit the calculations for the other days. Here is the result for all days:
Day
0
1
2
stock
$40.00
$40.50
$39.25
put
$199.05
$178.08
$230.55
delta
0.417596
0.385797
0.46892
Investment
1, 869.43 $1, 740.56 $2, 071.07
Interest credited (end of the day)
$0.41
$0.381 5
$0.45
Capital gain (end of the day)
$0.09
$4. 245 4
$0.05
Daily profit (end of the day)
$0.50
$3. 864
$0.40
www.actuary88.com
c
Yufeng
Guo
136
3
$38.75
$254.05
0.50436
$2, 208.46
$0.48
$4.48
$4.00
4
$40.00
$195.49
0.41940
$1, 873.10
$0.41
$0.91
$1.32
5
$40.00
$194.58
0.41986
$1, 874.02
Day
stock
put
delta
Investment
Interest credited (end of the day)
Capital gain (end of the day)
Daily profit (end of the day)
0
$40.00
$199.05
0.41760
$1, 869.43
$0.41
$0.42
$0.01
1
$40.642
172.6644
0.37684
$1, 704.22
$0.37
$0.35
$0.02
2
40.018
196.5319
0.41731
$1, 866.51
$0.41
$0.40
$0.01
Day
stock
put
delta
Investment
Interest credited (end of the day)
Capital gain (end of the day)
Daily profit (end of the day)
3
39.403
222.5962
0.45918
$2, 031.89
$0.45
$0.45
$0.00
4
$38.80
250.8701
0.50202
$2, 198.56
$0.48
$0.48
$0.00
5
39.420
220.0727
0.45940
$2, 031.02
Problem 13.6.
Problem 13.7.
If SOA tests this type of problems in the exam, theyll need to give you at
least and . You can calculate the option premium V and delta . Once you
have Greeks, just use the formula:
1
V (St+h , T t h) V (St , T t) + t + h + t
2
(Textbook 13.6)
Im not going to do all the parts in the problem. Im just going to show you
some examples.
First, youll need to get Greeks. You can use the BlackScholes formula and
find the option premium V and delta . To find and , youll need to use the
spreadsheet attached to the textbook.
= 0.3
r = 8%
=0
K = 40
www.actuary88.com
c
Yufeng
Guo
137
0)
=
4.1217
r = 8%
Day
time t
Expiration T t
stock price St
Call price Vt
Delta t
Gamma t
Theta t
=0
K = 40
0
0
180/365 = 0.493 150 684 9
40
4.1217
0.6151
0.0454
0.0134
5
h = 5/365
175/365 = 0.479 452 054 8
44
6.8440
0.7726
0.0330
0.0138
c
Yufeng
Guo
138
1
h = 1/365
179/365 = 0.490 410 958 9
44
1.3602
0.2280
0.0327
0.0053
r = 8%
=0
K = 40
Day
0
5
time t
0
h = 5/365
Expiration T t 180/365 = 0.493 150 684 9 175/365 = 0.479 452 054 8
Stock price St
40
44
Put price Vt
2.5744
1.3388
Delta t
0.3849
0.2274
Gamma t
0.0454
0.0330
Theta t
0.0050
0.0054
V (St , T t) = V (40, 180/365
0)
=
2.5744
c
Yufeng
Guo
139
S = 40
r = 8%
=0
K = 40
Day
0
1
time t
0
h = 1/365
Expiration T t 91/365
90/365
Stock price St
40
30
Call price Vt
2.7804
0.0730
Delta t
0.5824
0.0423
Gamma t
0.0652
0.0202
Theta t
0.0173 0.9134
a. The price of a 40strike, 90 day to expiration call is worth 0.0730
b. Use delta approximation:
= 30 40 = 10
0
V = V0 + t = 2.7804 + 0.5824 (10) = 3. 043 6
We get a nonsense value of 3. 043 6. This is because the delta approximation is good when is small. Here we have a large change of = 10.
c. Use deltagamma approximation:
1
0
V = V0 + t + t 2
2
1
= 2.7804 + 0.5824 (10) + 0.0652 (10)2 = 0.216 4
2
d. Use deltagammatheta approximation:
1
0
V = V0 + t + t 2 + t h
2
1
1
2
= 2.7804 + 0.5824 (10) + 0.0652 (10) 0.0173 365
2
365
= 0.199 1
We see in a, b, c, and d, the approximation is not good. This is because the
approximation is good when is small. Here we have a large change of = 10.
Problem 13.10.
www.actuary88.com
c
Yufeng
Guo
140
S = 40
r = 8%
=0
K = 40
Day
0
1
time t
0
h = 1/365
Expiration T t 1
364/365
Stock price St
40
30
Call price Vt
6.2845
6.9504
Delta t
0.6615
0.6909
Gamma t
0.0305
0.0287
Theta t
0.0104 0.0106
a. The price of a 40strike, 364 day to expiration call is worth 6.9504
b. Use delta approximation:
= 41 40 = 1
0
V = V0 + t = 6.2845 + 0.6615 (1) = 6. 946
c. Use deltagamma approximation:
1
0
V = V0 + t + t 2
2
1
= 6.2845 + 0.6615 (1) + 0.0305 (1)2 = 6. 961 3
2
d. Use deltagammatheta approximation:
1
0
V = V0 + t + t 2 + t h
2
1
1
= 6.2845 + 0.6615 (1) + 0.0305 (1)2 0.0104 365
= 6. 950 9
2
365
We see in a, b, c, and d, the approximation is good. This is because is
small.
Problem 13.11.
Im going to do one set of calculation assuming the stock price one day later
is $30.
= 0.3
S = 40
r = 8%
Day
0
time t
0
Expiration T t 91/365
Stock price St
40
Put price Vt
1.9905
Delta t
0.4176
Gamma t
0.0652
Theta t
0.0088
a. The price of a 40strike, 90
www.actuary88.com
=0
K = 40
1
h = 1/365
90/365
30
9.2917
0.9577
0.0202
0.0061
day to expiration put is worth 9.2917
c
Yufeng
Guo
141
S = 40
r = 8%
=0
K = 40
Day
0
1
time t
0
h = 1/365
Expiration T t 1
364/365
Stock price St
40
41
Put price Vt
3.2092
2.8831
Delta t
0.3385 0.3091
Gamma t
0.0305
0.0287
Theta t
0.0023 0.0025
a. The price of a 40strike, 364 day to expiration put is worth 2.8831
b. Use delta approximation:
= 41 40 = 1
0
V = V0 + t = 3.2092 0.3385 (1) = 2. 870 7
c. Use deltagamma approximation:
1
0
V = V0 + t + t 2
2
1
2
= 3.2092 0.3385 (1) + 0.0305 (1) = 2. 885 95
2
d. Use deltagammatheta approximation:
1
0
V = V0 + t + t 2 + t h
2
1
1
= 3.2092 0.3385 (1) + 0.0305 (1)2 0.0023 365
2
365
= 2. 883 65
Problem 13.13.
www.actuary88.com
c
Yufeng
Guo
142
1
2
2
13.9 is:
1
1
= 5. 29 106 0
365
Problem 13.15.
www.actuary88.com
c
Yufeng
Guo
143
40
45
30%
8%
= 180/365
0%
45strike call
2.1004
0.3949
0.0457
We bought one 45strike call. Suppose we need to sell (i.e. write) X unit
of 40strike call. The Gamma of the bought 45strike call is 0.0457 (negative
because we bought a call). The Gamma of the sold X unit of 40strike call is
0.0454X. The total Gamma of our portfolio is 0.0454X 0.0457. To Gamma
hedge, set 0.0454X 0.0457 = 0. This gives us X = 1. 006 6. The total Delta
of our portfolio is 0.6151 1. 006 6 0.3949 = 0.224 3. To delta hedge, we need
to buy 0.224 3 share of the underlying stock.
This is our final portfolio at time zero:
Transactions at time zero
Cost
Buy a 45strike call
2.1004
Sell 1. 006 6 unit of 40strike call (to Gamma hedge) 1. 006 6 4.1217 = 4. 148 9
Buy 0.224 3 share of the stock (to Delta hedge)
0.224 3 40 = 8. 972
Borrow 2.1004 4. 148 9 + 8. 972 = 6. 923 5
6. 923 5
Total
2.1004 4. 148 9 + 8. 972 6. 923 5 = 0
One day later, you close out your position:
Transactions (one day later)
Revenue
0
Sell a 45strike call
C1
0
Buy 1. 006 6 unit of 40strike call 1. 006 6C2
0
Sell 0.224 3 share of the stock
0.224 3S
Repay the borrowed amount
6. 923 5e0.08/365
0
0
0
Total
C1 1. 006 6C2 + 0.224 3S 6. 923 5e0.08/365
0
c
Yufeng
Guo
144
assume S = 50. Also assume that the stock volatility and risk free interest
rate are not changed. Using the BlackScholes formula (use T = 179/365), we
get:
0
0
C1 = 8.1511
C2 = 12.0043
The overnight profit is:
8.1511 1. 006 6 12.0043 + 0.224 3 50 6. 923 5e0.08/365 = 0.36
0
By changing S , youll get dierent profits. Then you draw a graph on how
0
the overnight profit varies by S . Since its timeconsuming to calculate the
0
overnight profit by changing S , Im not going to do it.
Problem 13.16.
Use DMs spreadsheet. The inputs are:
Stock Price
Exercise Price
Volatility
Riskfree interest rate
Time to Expiration (years)
Dividend Yield
40
45
30%
8%
= 180/365
0%
c
Yufeng
Guo
Revenue
5.3596
1. 006 6 4.1217 = 4. 148 9
1. 224 3 40 = 48. 972
50. 182 7
5.3596 4. 148 9 + 48. 972 50. 182 7 = 0
145
r = 8%
=0
T = 180/365
Suppose we sell X units of 180 day to expiration call. The total Gamma of
the written butterfly spread and written X units of 180 day to expiration call
is 0.037 7 + 0.0454X.
To Gamma hedge, set 0.037 7 + 0.0454X = 0, X = 0.830 4. The total
Delta of the written butterfly and the written X units of 180 day to expiration
call is 0.019 1 + 0.830 4 0.6151 = 0.491 7. So we need to buy 0.491 7 share of
the underlying stock.
Transactions at time zero
Sell the spread and buy 0.491 7 stock
Borrow 18. 126 3 from a bank
Total
www.actuary88.com
c
Yufeng
Guo
146
Revenue
0
C1
0
2C2
0
C3
0
0.491 7S
18. 126 3e0.08/365
0
0
0
0
C1 + 2C2 C3 + 0.491 7S 18. 126 3e0.08/365
0
r = 8%
=0
T = 180/365
c
Yufeng
Guo
Revenue
1. 101 4 + 1. 632 2 40 = 66. 389 4
66. 389 4
0
Revenue
0
2P1
0
P2
0
1. 632 2S
66. 389 4e0.08/365
0
0
0
2P1 + P2 1. 632 2S + 66. 389 4e0.08/365
147
www.actuary88.com
c
Yufeng
Guo
148
Chapter 14
Exotic options: I
Problem 14.1.
For n nonnegative numbers (such as stock prices) S1 , S2 , ..., Sn , the arithmetic mean can never be less than the geometric mean:
S1 + S2 + ... + Sn
1/n
(S1 S2 ...Sn )
n
If and only if S1 = S2 = ... = Sn , we have:
S1 + S2 + ... + Sn
= (S1 S2 ...Sn )1/n
n
The proof can be found at Wikipedia:
http://en.wikipedia.org/wiki/Inequality_of_arithmetic_and_geometric_
means
For example, this is the proof for n = 2.
(S1 S2 )2 0
S12 + S22 2S1 S2
2
(S1 + S2 ) = S12 + S22 + 2S1 S2 4S1 S2
2
S1 + S2
S1 + S2
S1 S2
S1 S2
2
2
Problem 14.2.
Arithmetic average:
5+4+5+6+5
= 5.0
5
Geometric average:
(5 4 5 6 5)1/5 = 4. 959
Ignore the question "What happens to the dierence between the two measures of the averages as the standard deviation of the observations increase?"
This question is vague. Im not sure what the author is after.
149
t = 0.5
t=1
Suu = 165.57665
Su = 128.67659
S = 100
Sud = 108.32871
Sd = 84.18680
Sdd = 70.87417
128.67659 + 108.32871
= 118. 502 65
2
(ud)
84.18680 + 108.32871
= 96. 257 8
2
84.18680 + 70.87417
= 77. 530 485
2
Path
uu
ud
du
dd
Average asset
147. 126 62
118. 502 65
96. 257 8
77. 530 485
Call Payo
47. 126 62
18. 502 65
0
0
(du)
(dd)
Risk Neutral Prob
2u
u d
u d
2d
c
Yufeng
Guo
150
(uu)
(ud)
(du)
Average asset
145. 965 2
118. 065 1
95. 497 9
77. 244 2
(dd)
Call Payo
45. 965 2
18. 065 1
0
0
average strike
K
147. 126 62
118. 502 65
96. 257 8
77. 530 485
call:
Call Payo
18. 450 03
0
12. 070 91
0
c
Yufeng
Guo
151
128.814749
108.32871
84.18680
91.10067
70.87417
64.42843
www.actuary88.com
c
Yufeng
Guo
152
c
Yufeng
Guo
153
Problem 14.7.
DM Chapter 14 doesnt have any formula on the price of a barrier option. To
calculate the barrier option price, you have to use the textbooks spreadsheet.
So this problem is out of the scope of Exam MFE. I used the textbooks
Excel spreadsheet and calculated the following:
T
0.25
0.5
1
2
3
4
5
100
BS
0.9744
2.1304
4.1293
7.4398
10.2365
12.6969
14.9010
39.9861
DO
0.7323
1.2482
1.8217
2.4505
2.8529
3.1559
3.4003
5.3112
BS/DO
1.3306
1.7067
2.2667
3.0360
3.5881
4.0232
4.3823
7.5286
Problem 14.8.
Out of the scope of MFE. However, I used the textbook spreadsheet and
found the following:
www.actuary88.com
c
Yufeng
Guo
154
Problem 14.9.
I used the spreadsheet and calculated the following (T is expressed in months):
T
BS
UO
BS/UO
1
0.1727
0.1727
1.0003
2
0.5641
0.5479
1.0296
3
0.9744
0.8546
1.1401
4
1.3741
1.0384
1.3234
5
1.7593
1.1243
1.5649
6
2.1304
1.1468
1.8577
7
2.4886
1.1316
2.1991
8
2.8353
1.0954
2.5885
9
3.1718
1.0482
3.0260
10
3.4991
0.9962
3.5124
11
3.8180
0.9430
4.0488
12
4.1293
0.8906
4.6365
As T goes up, the BS/UO ratio goes up too.
Problem 14.10.
With K = 0.9, the standard put is worth 0.0188. With barrier 1 or 1.05, the
upandout barrier is also worth 0.0188. Why? DM page 452 and 453 have an
explanation. Here is the main point. If the exchange rate never hits 1, then the
standard put and the upandout put with 1 or 1.5 barrier have the same value.
The only way that the standard put is more valuable than the upandout put
is when the exchange rate goes up from 0.9 to 1 or to 1.05 and then falls below
0.9, in which case the upandout is dead yet the standard put has a positive
payo of K ST = 0.9 ST . However, such a scenario is rare and 0.9 ST is
www.actuary88.com
c
Yufeng
Guo
155
S = 31.723.
In other words, if the stock price at t1 = 1 is 31.723, then a call option
written at t1 = 1 and expires at T = 2 is exactly worth $2.In this case, well not
exercise the compound option. Well let the compound option expire worthless.
If the stock price at t1 = 1 is less than 31.723, then the call written at t1 = 1
and expires at T = 2 is worth less than $2 and well not exercise the compound
option. Well let the compound option expire worthless.
Only if the stock price at t1 = 1 is greater than 31.723 should we exercise
the compound option.
c. To find the price of the compound call (i.e. call on call), well use worksheet
called "Compound" provided by the DM textbook.
The inputs are:
Stock Price
40
Exercise Price to buy asset
40
Exercise Price to buy option
2
Volatility
30%
Riskfree interest rate
8%
Expiration for Option on Option (years)
1
Expiration for Underlying Option (years)
2
Dividend Yield
0%
c
Yufeng
Guo
156
31.723
44.3494
c
Yufeng
Guo
157
1
S
+ r + 2 T
K2
2
d1 =
d2 = d1 T
For foreign currency, = r , r = r$ , and S = x0 .
ln
d1 =
= 0.309 7
0.435 89 1
c
Yufeng
Guo
158
www.actuary88.com
c
Yufeng
Guo
159
www.actuary88.com
c
Yufeng
Guo
160
Chapter 18
Lognormal distribution
Problem 18.1.
x
Using DM 18.4 z =
, we find the equivalent draws from the a standard
= 0.258 2
= 0.774 6
= 1. 2910
15
15
15
2 (8)
= 2. 5820
15
15 (8)
= 1. 807 4
15
Problem 18.2.
x
, we get x = + z. The equivalent draws are:
Using DM 18.4 z =
0.8 + (1.7) 25 = 7. 7
0.8 + (0.55)
25 = 3. 55
0.8 + (0.3) 25 = 0.7
0.8 + (0.02) 25 = 0.7
0.8 + (0.85) 25 = 5. 05
Problem 18.3.
Linear combination of normal random variables is also normal.
x1 + x2 is normal. Its mean is E (x1 + x2 ) = E (x1 ) + E (x2 ) = 1 2 = 1.
Its variance is
V ar (x1 + x2 ) = V ar (x1 ) + V ar (x2 ) + 2Cov (x1 , x2 ) = 5 + 2 + 2 (1.3) = 9. 6
x1 x2 is normal. Its mean is E (x1 x2 ) = E (x1 ) E (x2 ) = 1 (2) = 3.
Its variance is
V ar (x1 x2 ) = V ar (x1 ) + V ar (x2 ) 2Cov (x1 , x2 ) = 5 + 2 2 (1.3) = 4. 4
Problem 18.4.
161
c
Yufeng
Guo
162
Problem 18.7.
Theres a typo in the problem. The table should say "Month" instead of
"Day" (i.e. should be Month 0, Month 1, ...,Day 4, not Day 0, Day 1, ...,Day
4).
This is similar to DM Table 11.1.
a. Stock A:
St
2
Month Stock price
rt =ln
rt r
St1
0
100
105
1
105
ln
= 0.04879 016 0.00238048
100
102
2
102
ln
= 0.02898 75 0.00084028
105
97
3
97 ln
= 0.05026 183 0.00252625
102
100
4
100
ln
= 0.03 045 921 0.00092776
97
Total
0.00000000 0.00667477
4
P
rt
0
mean monthly continuously compounded return: r =
= =0
4
4
v
uP
2
u 4
r
u
rt r
t t=1
0.00667477
=
= 4. 716 909 3
monthly standard deviation:
41
3
t=1
102
c
Yufeng
Guo
163
Month
Stock price
0
1
2
3
4
Total
100
105
102
97
100
rt =ln
St
St1
0.04879016
0.35667494
0.43592432
0.00000000
0
rt r
0.00238048
0.12721702
0.19003001
0.00092776
0.32055527
4
P
rt
0
2
u 4
r
u
rt r
t t=1
0.32055527
monthly standard deviation:
=
= 0.326 881 87
41
3
4
P
rt
r = t=1
4
100
ln
100 = 0
4
4
P
t=1
4
P
St
S4
(ln St ln St1 )
ln
ln S4 ln S0
St1
S0
t=1
=
=
=
=
4
4
4
4
ln
2
u 4
u
rt r
t t=1
dard deviation in the following formula:
.
41
Problem 18.8.
ln St N ln S0 + 0.5 2 t, 2 t
Then for a given stock price c
P (St c) = P (ln St ln c) =
www.actuary88.com
!
ln c ln S0 0.5 2 t
c
Yufeng
Guo
164
ln c 4. 605 17 0.035 t
0.3 t
For t = 1, we have:
ln c 4. 640 17
0.3
ln 105 4. 640 17
0.3
ln c ln S0 0.5 2 t
ln c ln S0 0.5 2 t
0.5 2
0.5 2 d
d ln c ln S0 0.5 2 t
d
t=
=
z=
dt
dt
dt
t
2 t
For this problem, = 0.08, = 0, = 0.3
d
0.005 833
0.08 0 0.5 0.32
=
z=
dt
2 0.3 t
t
So z is a decreasing function of t. Remember that the accumulative normal
distribution (z) is an increasing function of z, we conclude:
The higher the t, the lower the z, the lower the P (St c) = (z), and the
higher the P (St > c) = 1 (z).
Similarly,
www.actuary88.com
c
Yufeng
Guo
165
d 0.5 2
d 0.5 2 t
d
d ln c ln S0 0.5 2 t
=
= t
z=
d
d
d
d
t
t
= t
0.5 = t 0.5 +
d
2
For this problem, = 0.08, = 0 so
0.08
d
> 0.
z = t 0.5 + 2
d
So z is an increasing function of .
The higher the , the higher the z, the higher the P (St c) = (z), and
the lower the P (St > c) = 1 (z).
Problem 18.9.
Use DM 18.30.
N d1
E (St St > K) = Se()t
N d2
d1
ln
=
S0
+ + 0.5 2 t
K
d2
= d1 T
In this problem,
1
Bien plus que des documents.
Découvrez tout ce que Scribd a à offrir, dont les livres et les livres audio des principaux éditeurs.
Annulez à tout moment.