1.

Put-call parity
C(K, T) P(K, T) = e
rT
(F
0,T
K) = F
P
0
Ke
rT
Continuous: F
P
0,T
= S
0
e
T
, so C(K, T) P(K, T) = S
0
e
T
Ke
rT
.
Discrete: F
P
0,T
= S
0

_
d
t
e
rt
= S
0
PV(dividends), so C(K, T) P(K, T) = S
0

_
d
t
e
rt
Ke
rT
.
To create one unit of synthetic stock: S
0
=
C(S,K)P(S,K)
e
T
+
Ke
rT
e
T
, so buy
1
e
T
calls, sell
1
e
T
puts, buy
Ke
rT
e
T
bonds.
2. Comparing options
For a Call: higher strike = lower price. For a Put: higher strike = higher price. 1
C
K
0 and 0
P
K
1.
American options are worth more than European.
S C
Amer
(S, K, T) C
Eur
(S, K, T) max(0, F
P
0,T
Ke
rT
)
K P
Amer
(S, K, T) P
Eur
(S, K, T) max(0, Ke
rT
F
P
0,T
)
The only rational time to exercise an American call option early is just before a dividend. If no dividends, then never exercise early.
Early Call exercise may be rational if PV
t,T
(Div) K(1 e
r(Tt)
). Early Put exercise may be rational if Call + PV
t,T
(Div) K(1 e
r(Tt)
).
An American option with exercise time T is one with exercise time t < T.
A European call option on a non-dividend stock with exercise time T is one with exercise time t < T.
A European option on a non-dividend stock with strike price Ke
r(Tt)
is one with strike price K.
3. Binomial trees stock, one period
Option payo at upper, lower nodes: C
u
, C
d
. Must have d < F = S e
(r)h
< u to prevent arbitrage.
Replicating portfolio: solving S u + e
r
B = C
u
and S d + e
r
B = C
d
gives
# shares to buy: =
C
u
C
d
S (u d)
e
h
, # bonds to buy: B =
uC
d
dC
u
u d
e
rh
, risk-neutral probability: p

=
e
(r)h
d
u d
These formulas can also be used to pull back the replicating portfolio (, B) at each node.
For a tree built using forward rates: u/d = e
(r)h

h
. Option price: C = S + B = e
rh
(p

C
u
+ (1 p

)C
d
)
To pull back to previous node, discount with e
rh
, not with the e
(r)h
that appears in the expression for p

!
4. Binomial trees general
For European trees, use binomial theorem shortcut:
_
n
k=0
_
n
k
_
p
k
(1 p)
nk
.
For American options, compare pulled-back value to exercise value (S
node
K) 0 or (K S
node
) 0.
For currency options, use = r
foreign
.
Futures: p

=
1 d
u d
, =
C
u
C
d
F
u
F
d
, B = C = e
rh
(P

C
u
+ (1 P

)C
d
For computing the forward price, F = S e
(r)T
where T is the length of the contract (nothing to do with the period h of the tree).
Stock/index Currency Futures Commodity Bond
dividend yield r
foreign
0 lease rate coupon yield
risk-free rate r r
domestic
r r r
5. Risk-neutral pricing
5.1. Pricing with true probabilities. is the expected return on a stock, and is the corresponding discounting rate for the option > r.
is also called the compound annual return for the option. For puts, usually < 0.
p

is not really the probability that the stock increase in value. It is the probability that makes = r.
p =
e
()h
d
u d
e
h
=
S
S + B
e
h
+
B
S + B
e
rh
Ce
h
= S e
h
+ Be
rh
= pC
u
+ (1 p)C
d
5.2. Risk-neutral pricing. U
i
is PV(1) for stock in state i. Q
i
is conditional value of PV(1), so Q
u
= pU
u
and Q
d
= (1p)U
d
and Q
u
+ Q
d
=
1
1 + r
.
Current value of a stock: C
0
= Q
u
C
u
+ Q
d
C
d
= pU
u
C
u
+ (1 p)U
d
C
d
If risk-neutral, Q
u
=
p
1 + r
, Q
d
=
1 p
1 + r
Eective annual rate of return: 1 + =
pC
u
+ (1 p)C
d
pU
u
C
u
+ (1 p)U
d
C
d
=
pC
u
+ (1 p)C
d
Q
u
C
u
+ Q
d
C
d
Risk-neutral probabilities: p

=
Q
u
Q
u
+ Q
d
=
pU
u
pU
u
+ (1 p)U
d
= Q
u
(1 + r) 1 p

=
Q
d
Q
u
+ Q
d
1
2
6. Binomial trees: miscellany
For early exercise to be optimal: div int put, or S (1 e
t
) K(1 e
rt
) Put(S, K)
Standard tree centered on (r )h, so u/d = e
(r)h

h
. Based on forward prices.
Cox-Ross-Rubenstein tree centered on 1, so u/d = e

h
. Allows arbitrage i e

h
< e
(r)h
.
Lognormal tree centered on (r

2
2
)h, so u/d = e
(r

2
2
)h

h
. Allows arbitrage i e
(r

2
2
)h+

h
< e
(r)h
.
Also called Jarrow-Rudd. Makes p very close to 0.5.
Estimating volatility: given stock prices S
t
i
for i = 0, 1, . . . , n, compute
i
:= ln(S
t
i
/S
t
i1
), for i = 1, . . . , n. Then
x =
1
n
n

i=1

i
and
2
=
1
n
n

i=1

2
i
and s
2
=
n
n 1
(
2
x
2
) unbiased annualized variance = s/

h
7. Modeling stock prices with the lognormal distribution
Weaknesses/assumptions: (i) constant volatility, (ii) stock returns for dierent periods are independent, (iii) stock prices dont jump.
For S
t
lognormal, ln(S
t
/S
0
) A
_
(

2
2
)t,
2
t
_
, so the annual continuously compounded mean return is

2
2
.
Pr(S
t
> K) = N(

d
2
) and Pr(S
t
< K) = N(

d
2
) , where

d
2
has in place of r:

d
2
=
ln(S
0
/K) + (

2
2
)t

t
Median stock price = S
0
e
(

2
2
)t
= E(S
t
)e

2
2
)t
< E(S
t
) = S
0
e
()t
= mean stock price.
To have S
L
S
t
S
U
with probability p: S
L
= S
0
e
(

2
2
)t+

tN
1
(p/2)
and S
U
= S
0
e
(

2
2
)t+

tN
1
(1p/2)
Conditional expected stock price: E[S
t
S
t
> K] = S e
()t
N(

d
1
)
N(

d
2
)
and E[S
t
S
t
< K] = S e
()t
N(

d
1
)
N(

d
2
)
Expected payo on a Call: E[max0, S
t
K] = S e
()t
N(

d
1
) KN(

d
2
) Put: E[max0, K S
t
] = KN(

d
2
) S e
()t
N(

d
1
)
C = e
rt
_

K
(S
t
K)g

(S
t
) dS
t
= e
rt
E

[S KS > K]P

[S > K] = S e
(r)t
N(

d
1
)
N(

d
2
)
N(d
2
) Ke
rt
N(

d
1
)
N(

d
2
)
N(d
2
) = B-S, for = r.
8. Fitting stock prices to a lognormal distribution
Estimate (with dividends removed) with
= +

2
2
= p x =

p
_
n
n 1
_
1
n

x
2
i
x
2
_
where n is number of observations and p is number of periods per year and x
i
= ln(S
i
/S
i1
)
9. The Black-Scholes formula
Assumptions:
Continuously compounded returns on the stock () are normally distributed and independent over time.
Continuously compounded returns on the strike asset (r) are known and constant.
Volatility is known and constant.
Dividends are known and constant.
There are no transaction costs or taxes.
It is possible to short-sell any amount of stock and to borrow any amount of money at the risk-free rate.
General Black-Scholes C(S, K, , T, r, ) = F
P
(S )N(d
1
) F
P
(K)N(d
2
) , where d
1
=
ln
_
F
P
(S )/F
P
(K)
_
+

2
2
T

T
and d
2
= d
1

T .
For stocks, F
P
(S ) = S e
T
and F
P
(K) = Ke
rT
, so d
1
=
1

T
_
ln
_
S
K
_
+
_
r +

2
2
_
T
_
and d
2
= d
1

T and
Call(S, K) = S e
T
N(d
1
) Ke
rT
N(d
2
), Put(S, K) = Ke
rT
N(d
2
) S e
T
N(d
1
).
For currency, let r
f
= r
foreign
and let x be the amount of foreign currency to buy. Then = r
f
and
Call(x, K) = xe
r
f
T
N(d
1
) Ke
rT
N(d
2
), Put(x, K) = Ke
rT
N(d
2
) se
r
f
T
N(d
1
).
For futures, = r, so d
1
=
1

T
_
ln
_
S
K
_
+

2
2
T
_
. When K = S , this becomes d
1
=

T/2. The Black formula is:

Call(x, K) = Fe
rT
N(d
1
) Ke
rT
N(d
2
), Put(x, K) = Ke
rT
N(d
2
) Fe
rT
N(d
1
).
3
10. The Black-Scholes formula: Greeks

put
=
call
e
T
S-shaped curve

put
=
call
symmetric hump, peak to left of stock price, further left with higher T
vega vega
put
= vega
call
asymmetric hump, peak like

put
=
call
+
1
365
(rKe
rT
S e
T
) Sulcus for short lives, gradual decrease for long lives.

put
=
call

T
100
Ke
rT
increasing curve (pos for calls, neg for puts)

put
= +
T
100
S e
T

call
decreasing curve (neg for calls, pos for puts)
To convert between Put and Call greeks, dierentiate both sides of C P = S e
t
Ke
rt
with respect to the appropriate variable.
is almost always negative for calls; negative for puts unless far in the money.
Acall option can be replicated by buying shares of stock and borrowing Ke
rT
N(d
2
).
call
= e
T
N(d
1
) and
put
= e
T
N(d
1
) =
call
e
T
Elasticity of the option with premium C is =
S
C
=
/C
/S
=
%option change
%stock change
= percentage risk. Dollar risk is .
For calls, 1 because 0 and S = S e
T
N(d
1
) > C(S ). For puts, 0 because 0.
Volatility of the option is
option
=
stock
.
Sharpe ratio is
r

option
=
r

stock
. r = ( r) is true instantaneously, and follows from e
h
= e
rh
+ (e
h
e
rh
).
For a portfolio consisting of a
i
of options C
i
: A greek for the porfolio is computed by
portfolio
=

a
i

C
i
.
Elasticity for the porfolio is computed by
portfolio
=
S
portfolio
C
portfolio
=
S
_
a
i

C
i
_
a
i
C
i
.
11. The Black-Scholes formula: applications, implied volatility
Calendar spread Sell C(S, K, t) and buy C(S, K, T), with t < T (or buy C(S, K, t) and sell C(S, K, T)).
If C
t
<< C
0
, then loss on C(S, K, T) outweighs prot on sale of C(S, K, t).
If C
t
>> C
0
, then obligation from sale of C(S, K, t) outweighs prot when exercising C(S, K, T).
To nd holding period prot/calendar spread prot for [0, t], where 0 < t < T
i
, use BS to compute
(portfolio value at time t) (portfolio value at time 0)e
rt
=
n

i=1
C
i
(S, K, T
i
) + (cash settlements)
m

j=1
C
j
(S, K, t)e
rt
. (1)
Implied volatility (i) Allows pricing other options on the same stock, without market prices. (ii) Is a quick way to describe option prices. (iii)
Volatility skew measures accuracy of Black-Scholes model. Volatility skew: implied volatility tends to be lower for high strike prices.
12. Delta hedging
If a market-maker sells an option, he buys of the stock to hedge so there will be no prot or loss if the stock price changes.
12.1. Overnight prot. Ignoring dividends, prot is change in option value & (change in stock price) & interest on borrowed money:
market-maker prot = (C(S
t
) C(S
0
)) +(S
t
S
0
)
_
e
rt/365
1
_
(S
0
C(S
0
)), where t is in days (usually t = 1). This is a special case of (1):
market-maker prot = (S
t
C(S
t
)) e
rt/365
(S
0
C(S
0
)) which is positive i S

t < S
t
< S +

t .
12.2. Delta-gamma-theta approximation. C(S
t
+ ) = C(S ) + +
1
2

2
+ t + Taylor remainder error.
The whole point of -hedging is to separate (or

h) from . For =

h, get
market-maker prot =
_

2
2
+ rh(S C(S )) + h
_
=
_

2
2
S
2
+ r(S C(S )) +
_
h (2)
12.3. Rehedging. Buy
t

0
of the stock, where
t
= N(d
1
) is computed at time t, and
0
= N(d
1
) computed at time 0. If negative, sell.
Let R
h,i
be period-i return to a delta-hedged market maker who has written a call. Then
R
h,i
=

2
2
S
2
(x
2
i
1)h, and Var(R
h,i
) =
1
2
_

2
S
2
h
_
2
and annual variance =
1
2
_

2
S
2

_
2
h
12.4. Hedging multiple greeks. Sell option C
sell
, then - hedge by buying x
1
shares of stock and buying x
2
of another option C
buy
:
Delta-gamma hedging
stock C
buy
C
sell
: x
1
+
buy
x
2
=
sell
:
buy
x
2
=
sell
compare to Delta-hedging
stock C
sell
: x
1
=
sell
4
13. Asian and barrier options
13.1. Asian options. Arithmetic average A(S ) =
1
T
_
T
t=1
S
t
and geometric average G(S ) =
T
_

T
t=1
S
t
. Ignore initial price (exclude S
0
)
arithmetic geometric
average price Call = max(0, A(S ) K), Put = max(0, K A(S )) Call = max(0, G(S ) K), Put = max(0, K G(S ))
average strike Call = max(0, S
T
A(S )), Put = max(0, A(S ) S
T
) Call = max(0, S
T
G(S )), Put = max(0, G(S ) S
T
)
Asian is cheaper than European, since less volatile. Similarly, average over more items is cheaper.
Daily average price < monthly average price. Monthly average strike < daily average strike.
G(S ) A(S ) = Geometric average price call < arithmetic average price call. Reverse inequality for puts, also reverse for average strikes.
13.2. Barrier options. Rebate options: pay a xed amount when the barrier is hit.
Parity: knock-in option + knock-out option = ordinary option .
13.3. Compound options. For binomial tree models, work out the binomial tree for the underlying rst. Then for the compound option, work out
a second tree with initial vertices given by the prices of the underlying.
13.3.1. Compound option parity.
CallOnCall(C(S, K, T), x, t) PutOnCall(C(S, K, T), x, t) = C(S, K, T) xe
rt
CallOnPut(P(S, K, T), x, t) PutOnPut(P(S, K, T), x, t) = P(S, K, T) xe
rt
13.3.2. American options on stocks with one discrete dividend. If a dividend D is paid at time t, then value of an American call at time t is greater
of exercise value and option for remaining period: maxS
t
+ D K, C(S
t
, K, T t). If S
t
is cum-dividend and S

t
= S
t
D is ex-dividend, then
time t : S

t
+ D K + max0, P(S

t
, K, T t) + K(1 e
r(Tt)
) D = time 0 : S
0
Ke
rt
+ CallOnPut[S, K, D K(1 e
r(Tt)
)] .
Early exercise is not rational if x = D K(1 e
r(Tt)
) < P(S

t
, D K(1 e
r(Tt)
)) . For P, use F
P
(S ) = S
0
De
rt
and F
P
(K) = Ke
rT
.
14. Compound, gap, and exchange options
14.1. All-or-nothing options.
Option S S > K S S < K c S > K c S < K
Value S e
T
N(d
1
) S e
T
N(d
1
) Ke
rT
N(d
2
) Ke
rT
N(d
2
)
14.2. Gap options. K splits into K
strike
and K
trigger
. Occurrence of payo is determined by K
trigger
, so use it to determine the probabilities N(d
i
).
Amount of payo is determined by K
strike
, so use it to compute option price:
Call = S e
t
N(d
1
) K
strike
e
rt
N(d
2
), Put = K
strike
e
rt
N(d
2
) S e
t
N(d
1
).
where d
1
=
1

T
_
ln(S/K
trigger
) +
_
r +

2
2
_
T
_
and d
2
= d
1

T. Decompose gap in terms of AoNs:

C(S, K
strike
, K
trigger
) = C(S, K
trigger
) + (K
trigger
K
strike
)S > K
trigger
P(S, K
strike
, K
trigger
) = P(S, K
trigger
) + (K
trigger
K
strike
)S < K
trigger
14.3. Exchange options. Let S be the asset you receive, with dividend rate
1
, and K be the asset you may exchange for it, with dividend rate
2
.
The volatility of S K is
2
=
2
S
+
2
K
2
S

K
14.4. Chooser options. V = C(S, K, T) + e
(Tt)
P(S, Ke
(r)(Tt)
, t)
14.5. Forward-start options. Let d
Tt
i
be d
i
computed with time T t instead of t. Then
S
t
e
(Tt)
N(d
Tt
1
) S
t
e
r(Tt)
N(d
Tt
2
)
discount to t=0
S
0
e
T
N(d
Tt
1
) S
0
e
r(Tt)T
N(d
Tt
2
)
14.6. For hedging: dierentiate using
N(d
i
)
S
=

S
_
d
i
(S )

e
(x)
2
/2

2
dx =
e
(d
i
)
2
/2

2
_
d
i
S
_
=
e
(d
i
)
2
/2

2
_
1
S

T
_
=
e
(d
i
)
2
/2
S

2T
.
15. Monte Carlo valuation
To simulate a lognormal random variable, let u U(0, 1) be uniform. Then N
1
(u) A(0, 1) and e
N
1
(u)
is lognormal.
If V(S
T
, T) is option payo at T, the Monte Carlo time-0 price is V(S
0
, 0) =
e
rT
n
n

i=1
V(S
i
T
, T)
For a European call, this would be C =
e
rT
n
n

i=1
max
_
0, S
0
e
(r

2
2
)T+

TZ
i
K
_
, where Z
i
A(0, 1) for i = 1, . . . , n.
x is sample mean and
C
= s
n
=
_
1
n
_
n
i=1
(x
i
x)
2
_
1/2
is sample stdev for one draw. Then
n
=
s
n

n
is stdev of the Monte Carlo estimate.
To attain a given target standard error of
n
, need n = (s
n
/
n
)
2
trials.
5
16. Brownian motion
A
t
is stock price at time t, is continuous rate of return, is continuous dividend rate, is volatility, N(x) is the normal (cumulative) distribution
function. So total drift is =capital gains return=contin. compounded expected incr..
If or is given in terms of a time unit, use this to denominate time (e.g., = 2 per quarter = 1 year would be t = 4).
For stock to exceed a continuously compounded annual return (yield) of y means
S (t)
S (0)
e
yt
, or ln
_
S (t)
S (0)
_
yt.
16.1. Arithmetic BM:. X(t) = s + Z(t). The increment X(t + h) X(t) has mean = ( )h and var=
2
h, so
P[X(t + h) < A
t+h
X(t) = A
t
] = N
_
A
t+h
A
t

h
_
, dX(t) = dt + dZ(t)
16.2. Geometric BM:. Y(t) = e
X(t)
. The increment Y(t + h) Y(t) has mean = (

2
2
)h and var=
2
h, so
P[Y(t + h) < A
t+h
Y(t) = A
t
] = P
_
Y(t + h)
Y(t)
<
A
t+h
A
t
Y(t) = A
t
_
= N
_
ln A
t+h
ln A
t

h
_
, dY(t) = Y(t) dt + X(t)dZ(t)
To go from geometric to arithmetic, subtract

2
2
and replace X(t) with ln X(t) .
Covariance. For standard BM: (Z
s
, Z
t
) := Cov(Z
s
, Z
t
) = mins, t.
For X
t
= X
0
+ t + Z
t
, (X
s
, X
t
) =
2
(Z
s
, Z
t
). For X
t
= X
0
e
t+Zt
, (Z
s
, Z
t
) = X
2
0
e
(+

2
2
)(s+t)
(e

2
(Zs,Zt )
1).
17. Differentials
Watch for variance (
2
) given instead of volatility (). Var(ln S (t)S (0)) = Var(ln F
0,T
(S )) = Var(ln F
P
0,T
(S )) =
2
t.
ln X(t)X(0) N
_
X(0) + (

2
2
)t,
2
t
_
X(t) = X(0)e
(

2
2
)t+Z(t)
ln X(t) ln X(0) = (

2
2
)t + Z(t) d (ln X(t)) = (

2
2
) dt + dZ
18. It os lemma
dC = C
S
dS +
1
2
C
S S
(dS )
2
+ C
t
dt. If dS is arithmetic BM, (dS )
2
=
2
dt. Dont forget
2
!
(dt)
2
= 0, dt dZ
t
= 0, (dZ
t
)
2
= dt, dZ
t
dZ

t
= dt, where is correlation coecient.
Ornstein-Uhlenbeck process: dX
t
= ( X
t
) dt + dZ
t
, where is speed of reversion to mean .
19. The Black-Scholes equation
Set (2) equal to 0 to obtain:

2
2
C
S S
S
2
+ rC
S
S + C
t
= rC(S ) or

2
2
S
2
+ rS + = rC(S )
Use it to price a claim C or to determine the parameters of a derivative security.
20. Sharpe ratio
Express process:
dX
X
=
_
(t, X) (t, X)
_
dt + (t, X)dZ. Then =
(t, X) r
(t, X)
.
Risk-free portfolios: buy x
1
of
dS
1
(t)
S
1
(t)
=
1
dt +
1
dt and buy x
2
of
dS
2
(t)
S
2
(t)
=
2
dt +
2
dt, to solve either of:

1
S
1
(0)x
1
+
2
S
2
(0)x
2
=
_
S
1
(0)x
1
+ S
2
(0)x
2
_
r or
1
S
1
(0)x
1
+
2
S
2
(0)x
2
= 0.
21. Risk-neutral pricing and proportional portfolios
Risk-neutral process for stocks (Girsanovs theorem): let =
r

be the Sharpe ratio, so dZ = dZ + dt is arithmetic BM. Then

geometric BM
dS (t)
S (t)
= dt + dZ risk-neutral BM
dS (t)
S (t)
= r dt + dZ
Blended portfolio:
dW(t)
W(t)
=
_
+ (1 )r
_
dt + dZ. Then
W(t)
W(0)
=
_
S (t)
S (0)
_

e
_

W
+(1)(r+

2
2
_
t
22. Monomial securities S
a
The process is S
a
t
= S (0)
a
e
a(

2
2
)t+aZt
and its expected value is E
_
S (T)
a
_
= S (0)
a
e
a(+(a1)

2
2
)T
Forward price is F
0,T
_
S (T)
a
_
= S (0)
a
e
a(r+(a1)

2
2
)T
. Prepaid forward price is F
P
0,T
_
S (T)
a
_
= e
rT
S (0)
a
e
a(r+(a1)

2
2
)T
The It o process C = S
a
is given by
d(S
a
)
S
a
=
_
a( + (a 1)

2
2
)T
_
dt + adZ.
Sharpe ratios
r
a
=
r

show that C = S
a
earns = a( r) + r. Then ln F
P
0,T
(S
a
) = ln E[S
a
] relates the above formulas.
Use for S
a
t
and E[S (T)
a
]; use r for F
0,T
(S (T)
a
) and F
P
0,T
(S (T)
a
). For options on S
a
, use = a in BS .
6
23. Stochastic integration
For dierentiating an integral:

t
__
t
0
f (s)dZ
s
_
= f (t) dZ
t
. Dont forget the dZ
t
!
To solve dX
t
+ A(t)X
t
dt = B(t) dt, use integrating factor = e
_
A(t) dt
.
Quadratic variation: for [0, T], let t
i
=
i
n
T, so equally spaced increments. Then QV(X, 0, T) = lim
n
n

i=1
_
X(t
i
) X(t
i1
)
_
2
.
Ornstein-Uhlenbeck: dX
t
= ( X
t
)dt + dZ
t
X
t
= X
0
e
t
+ (1 e
t
) +
_
t
0
e
(st)
dZ
t
.
24. Binomial trees for interest rates
rates
r
uu
r
u
g
g
g
g
g
g
g
g
g
U
U
U
U
U
U
U
U
U
r

h
h
h
h
h
h
h
h
h
h
V
V
V
V
V
V
V
V
V
V
r
ud
r
du
r
d
W
W
W
W
W
W
W
W
W
i
i
i
i
i
i
i
i
i
r
dd
=
prices
e
ruu
e
ru
2
_
e
(ru+ruu)
+ e
(ru+r
ud
)
_
d d d d d d d
Y Y Y Y Y Y Y Y
e
r

Paths
_
e
rn
d d d d d d d
Z Z Z Z Z Z Z
e
r
ud
e
r
du
e
r
d
2
_
e
(r
d
+r
du
)
+ e
(r
d
+r
dd
)
_
Z Z Z Z Z Z Z
f f f f f f f f
e
r
dd
24.1. Generic (nonBDT) interest rate trees. Given a continuously compounded interest rate tree with entries r (or prices P(t, t + 1) = e
r
),
P(0, T) =

Paths
P()
_
n
e
rn
, where r
n
are the rates appearing in the path (3)
To obtain the continuously compounded yield to maturity, solve P(0, T) = e
rT
for r, where T =number of periods (columns).
To compute the premium of a call at time t, strike K: compute (P(0, T) K) 0 for column t. Then walk back by average & discount.
Prices at the end of the tree are just given by e
r
; for intermediate prices, use (3).
For American options: immediate exercise at a node with rate r is worth e
r
, so exercise a call early if e
r
K > (pulled-back value).
24.2. Black-Derman-Toy trees. Use P
j
(t 1, T) =
1
1 + R
j
(t 1, t)
1
2
_
P
j
(t, T) + P
j+1
(t, T)
_
to walk back, and P =
1
1 + R
to convert.
BDT tree R
2t
e
4

t
R
t
e
2

t
g g g g g g g
X X X X X X X
R
0
f
f
f
f
f
f
f
f
f
X X X X X X X X X X X
R
2t
e
2

t
R
t
Y Y Y Y Y Y Y Y Y Y
f
f
f
f
f
f
f
f
R
2t
Construct with:
P(0, 2)
P(0, 1)
=
1
2
_
1
1 + R
1
+
1
1 + R
1
e
2

t
_
Ratio in column t:
rate
j
rate
j+1
= e
2

t
To get volatility for yields, use =
1
2
ln
_
R
j
(t,T)
R
j+1
(t,T)
_
from column ratio. For t T 1, compute rates from P
j
(t, T)
To compute P(0, T), start with prices P
j
(T 1, T) =
1
1+R
j
(T1,T)
in column T. Then walk back by average & discount.
To compute standard F
standard
0,T
, start with discounted rates R
j
(T 1, T) =
R
1+R
in column T. Then walk back by average & discount.
R
standard
A
=
P(0, T)
P(0, T + 1)
1 =
P(0, T) P(0, T + 1)
P(0, T + 1)
=
F
standard
0,T
P(0, T + 1)
To compute Eurodollar-style F
Eurodollar
0,T
, start with rates R
j
(T 1, T) =
1
P
j
(T1,T)
1 in column T. Then walk back by average & discount.
Equivalently, start with discounted prices
1
1+R
j
(T2,T1)
1
2
_
R
j
(T 1, T) + R
j+1
(T 1, T)
_
in column T 1 and walk back by average & discount.
R
Eurodollar
A
=
F
Eurodollar
0,T
P(0, T)
The forward rate R
A
for an FRA (=interest rate for computing payo) is the rate that makes the expected value of the FRA equal to 0.
If R(t, T) is the rate of a loan starting at t and ending at T, then FRA pays R(t, T) R
A
at time T, or
_
1
1 + R(t, T)
_
Tt
(R(t, T) R
A
) at time t.
25. The Black formula
25.1. Bond options. Black formula for bond options: F
t
= F
0,t
[P(t, T)] =
P(0, T)
P(0, t)
is the price at t to purchase a bond maturing at T.
Call : C(F
t
, K) = P(0, t)[F
t
N(d
1
) K N(d
2
)]
Put : P(F
t
, K) = P(0, t)[K N(d
2
) F
t
N(d
1
)]
d
1
=
ln(F/K) +

2
2
t

t
, d
2
= d
1

t
(4)
7
25.2. Caps via the Black formula. Each caplet is (1 + K) puts with strike
1
1+K
. A Put with strike
1
1+K
and exercise time t has value P(F
t
,
1
1+K
).
Cap price = (1 + K)
_
P
0
+ P(F
1
,
1
1+K
) + + P(F
T
,
1
1+K
)
_
Caplet price = (1 + K)P(F
t
,
1
1+K
) from (4)
where P
0
= (
1
bond price at 0
1) 0 = R
0
0 is the initial payo, and P(F
t
,
1
1+K
) is computed using F
0,t
and
t
.
The strike K is constant throughout, but F
t
=
P(0, t)
P(0, t 1)
and t changes for each caplet . Remember to discount by P(0, t)!
26. Equilibrium interest rate models
26.1. The impossible model. Use continuously compounded interest P(0, T) = e
rT
.
Hedge ratio for duration-hedge: N =
t
1
P(0, t
1
)
t
2
P(0, t
2
)
. (< 0 = sell), N = N
t
2
=number of t
2
-bonds
26.2. Equilibrium models. P = P(r, T) is the price of a zero-coupon bond when the short-term rate is r, and dr = a(r) dt + (r) dZ.
dP = Pdt + qPdZ, = (r, t, T) =
1
P
(aP
r
+

2
2
P
rr
+ P
t
) q = q(r, t, T) =
1
P
P
r
The Sharpe ratio is
(r, t) =
(r, t, T) r
q(r, t, T)
=
1
P
r
_
aP
r
+

2
2
P
rr
+ P
t
rP
_
=
a

+
P
rr
2P
r
+
P
t
rP
P
r
Risk-neutral process for bonds: short-term interest rates dr = a dt + dZ risk-neutral interest rates dr = (a ) dt + dZ
Black-Scholes equation:

2
2
S
2
+ rS + = rC(S )

2
2
P
rr
+ (a )P
r
+ P
t
= rP
26.2.1. Rendelman-Bartter: dr = ar dt + r dZ . Advantages: r 0, vol r. Disadvantages: no mean reversion, r unbounded.
26.2.2. Vasicek: dr = a(b r) dt + dZ . Advantages: mean reversion. Disadvantages: vol is constant, can have r < 0.
P(r, t, T) = Ae
rB
, where A(t, T) = e
r[B(Tt)]B
2
2
4a , B = B(t, T) =
1
a
(1 e
a(Tt)
) , r = b +

a


2
2a
2
. (5)
Also, B(t, T) = a
Tt a
and r is the formula for the yield to maturity on an innitely-lived bond. The price model (5) satises

2
2
P
rr
+ (a(b r) )P
r
+ P
t
= rP, for P
r
= = BP and P
rr
= = B
2
P.
For the case a = 0: dr = dz, r is undened, B = T t, and A = exp(
B
2
2
+

2
2
B
3
3
).
26.2.3. Cox-Ingersoll-Ross: dr = a(b r) dt +

r dZ . Advantages: mean reversion, vol r, r 0.

P(r, t, T) = Ae
rB
, where A = A(t, T) =
_
2e
(a++)(Tt)/2
(a + + )(e
(Tt)
1) + 2
_
2ab/
2
, B = B(t, T) =
2(e
(Tt)
1)
(a + + )(e
(Tt)
1) + 2
,
and =
_
(a + )
2
+ 2
2
and = (r, t)(r, t)/r and r =
2ab
a + +
is yield to maturity on an innitely-lived bond.
26.2.4. Facts for Vasicek and CIR models.
Higher volatility = lower yield. Instantaneous rate of change = drift term of dr.
For CIR: higher risk premium = lower yield (same for Vasicek when a = 0).
A = A(t, T) = A(T t) and B = B(t, T) = B(T t) depend only on T t.
= P
r
= BP and = P
rr
= B
2
P , so q =
Pr
P
= q(r, t, T) = (r)B .
Vasicek: (r, t, T) = a(b r)B +

2
2
B
2
+
Pt
P
and the Sharpe ratio =
r
B
does not vary with r or t.
CIR: Sharpe ratio =

, so (r
1
, t) (r
2
, t) and (r, t)(r) =
_

_
_

r
_
= r, and

r
does not vary with r or t.
26.3. Delta-hedging.
duration-hedge :

(bond value) (bond duration) = 0 = need to sell N =

t
1
P(r, 0, t
1
)
t
2
P(r, 0, t
2
)
of bond 2
delta-hedge :

(bond value) (bond delta) = 0 = need to sell N =

P
r
(r, 0, t
1
)
P
r
(r, 0, t
2
)
of bond 2, P
r
=
P
r
8
2
0
4
0
6
0
8
0
0
0
.
2
0
.
4
0
.
6
0
.
8 1
S
t
o
c
k

p
r
i
c
e
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
D
e
l
t
a

f
o
r

a

c
a
l
l
2
0
4
0
6
0
8
0
0
1
0
2
0
3
0
4
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
C
a
l
l

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

p
r
i
c
e

S
S
t
o
c
k

p
r
i
c
e
D
2
0
4
0
6
0
8
0
0 5
1
0
1
5
2
0
2
5
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1
.
2

y
e
a
r
3
0

y
e
a
r
V
e
g
a

f
o
r

a

c
a
l
l

o
r

a

p
u
t
0
0
.
1
0
.
2
0
.
3
0
.
4
0
.
5
5
1
0
1
5
2
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
v
o
l
a
t
i
l
i
t
y
C
a
l
l

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

v
o
l
a
t
i
l
i
t
y

s
v
2
0
4
0
6
0
8
0

1
5

1
0

5
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
T
h
e
t
a

f
o
r

a

c
a
l
l
C
a
l
l

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

t
0
1
2
3
0 5
1
0
1
5
2
0
T
i
m
e

t
o

e
x
p
i
r
y
q
2
0
4
0
6
0
8
0
0
2
0
4
0
6
0
8
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
S
t
o
c
k

p
r
i
c
e
R
h
o

f
o
r

a

c
a
l
l
0
0
.
0
5
0
.
1
0
.
1
5
0
.
2
5
1
0
1
5
2
0
2
5
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
R
i
s
k

f
r
e
e

r
a
t
e

r
C
a
l
l

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

r
r
2
0
4
0
6
0
8
0

1
0
0

8
0

6
0

4
0

2
0
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
P
s
i

f
o
r

a

c
a
l
l
S
t
o
c
k

p
r
i
c
e
0
0
.
0
5
0
.
1
0
.
1
5
0
.
2
5
1
0
1
5
2
0
2
5
3
0
3
5
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
D
i
v
i
d
e
n
d

y
i
e
l
d

r
a
t
e

d
C
a
l
l

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

d
Y
2
0
4
0
6
0
8
0

0
.
8

0
.
6

0
.
4

0
.
2
0
D
e
l
t
a

f
o
r

a

p
u
t
S
t
o
c
k

p
r
i
c
e
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
2
0
4
0
6
0
8
0
0 5
1
0
1
5
2
0
2
5
3
0
P
u
t

p
r
e
m
i
u
m
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
S
t
o
c
k

p
r
i
c
e
P
u
t

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

p
r
i
c
e

S
D
2
0
4
0
6
0
8
0
D
e
l
t
a

f
o
r

a

c
a
l
l

o
r

a

p
u
t
S
t
o
c
k

p
r
i
c
e
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
2
0
4
0
6
0
8
0
0
0
.
0
2
0
.
0
4
0
.
0
6
0
.
0
8
G
a
m
m
a

f
o
r

a

c
a
l
l

o
r

a

p
u
t
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1
.
2

y
e
a
r
1
0

y
e
a
r
S
t
o
c
k

p
r
i
c
e
G
2
0
4
0
6
0
8
0

1
5

1
0

5
0 5
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
S
t
o
c
k

p
r
i
c
e
T
h
e
t
a

f
o
r

a

p
u
t
0
1
2
3
0 5
1
0
1
5
S
t
r
i
k
e

K
=
4
0
S
t
r
i
k
e

K
=
5
0
S
t
r
i
k
e

K
=
5
5
S
t
r
i
k
e

K
=
5
8
S
t
r
i
k
e

K
=
6
5
P
u
t

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

t
q
P
u
t

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

r
0
0
.
0
5
0
.
1
0
.
1
5
0
.
2
0 2 4 6 8
1
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
R
i
s
k

f
r
e
e

r
a
t
e

r
2
0
4
0
6
0
8
0

1
0
0

8
0

6
0

4
0

2
0
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
R
h
o

f
o
r

a

p
u
t
S
t
o
c
k

p
r
i
c
e
r
2
0
4
0
6
0
8
0
0
2
0
4
0
6
0
8
0
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
S
t
o
c
k

p
r
i
c
e
P
s
i

f
o
r

a

p
u
t
P
u
t

p
r
e
m
i
u
m

a
s

a

f
u
n
c
t
i
o
n

o
f

d
0
0
.
0
5
0
.
1
0
.
1
5
0
.
2
0
0
.
0
5
0
.
1
0
.
1
5
0
.
2
0
.
2
5
0
.
1

y
e
a
r
0
.
5

y
e
a
r
1

y
e
a
r
2

y
e
a
r
s
D
i
v
i
d
e
n
d

y
i
e
l
d

r
a
t
e

d
Y
S
t
r
i
k
e

K
=
4
0
S
t
r
i
k
e

K
=
5
0
S
t
r
i
k
e

K
=
5
5
S
t
r
i
k
e

K
=
6
5
S
t
r
i
k
e

K
=
8
0
Normal Distribution Table
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
0.0 0.50000 0.50399 0.50798 0.51197 0.51595 0.51994 0.52392 0.52790 0.53188 0.53586
0.1 0.53983 0.54380 0.54776 0.55172 0.55567 0.55962 0.56356 0.56749 0.57142 0.57535
0.2 0.57926 0.58317 0.58706 0.59095 0.59483 0.59871 0.60257 0.60642 0.61026 0.61409
0.3 0.61791 0.62172 0.62552 0.62930 0.63307 0.63683 0.64058 0.64431 0.64803 0.65173
0.4 0.65542 0.65910 0.66276 0.66640 0.67003 0.67364 0.67724 0.68082 0.68439 0.68793
0.5 0.69146 0.69497 0.69847 0.70194 0.70540 0.70884 0.71226 0.71566 0.71904 0.72240
0.6 0.72575 0.72907 0.73237 0.73565 0.73891 0.74215 0.74537 0.74857 0.75175 0.75490
0.7 0.75804 0.76115 0.76424 0.76730 0.77035 0.77337 0.77637 0.77935 0.78230 0.78524
0.8 0.78814 0.79103 0.79389 0.79673 0.79955 0.80234 0.80511 0.80785 0.81057 0.81327
0.9 0.81594 0.81859 0.82121 0.82381 0.82639 0.82894 0.83147 0.83398 0.83646 0.83891
1.0 0.84134 0.84375 0.84614 0.84849 0.85083 0.85314 0.85543 0.85769 0.85993 0.86214
1.1 0.86433 0.86650 0.86864 0.87076 0.87286 0.87493 0.87698 0.87900 0.88100 0.88298
1.2 0.88493 0.88686 0.88877 0.89065 0.89251 0.89435 0.89617 0.89796 0.89973 0.90147
1.3 0.90320 0.90490 0.90658 0.90824 0.90988 0.91149 0.91309 0.91466 0.91621 0.91774
1.4 0.91924 0.92073 0.92220 0.92364 0.92507 0.92647 0.92785 0.92922 0.93056 0.93189
1.5 0.93319 0.93448 0.93574 0.93699 0.93822 0.93943 0.94062 0.94179 0.94295 0.94408
1.6 0.94520 0.94630 0.94738 0.94845 0.94950 0.95053 0.95154 0.95254 0.95352 0.95449
1.7 0.95543 0.95637 0.95728 0.95818 0.95907 0.95994 0.96080 0.96164 0.96246 0.96327
1.8 0.96407 0.96485 0.96562 0.96638 0.96712 0.96784 0.96856 0.96926 0.96995 0.97062
1.9 0.97128 0.97193 0.97257 0.97320 0.97381 0.97441 0.97500 0.97558 0.97615 0.97670
2.0 0.97725 0.97778 0.97831 0.97882 0.97932 0.97982 0.98030 0.98077 0.98124 0.98169
2.1 0.98214 0.98257 0.98300 0.98341 0.98382 0.98422 0.98461 0.98500 0.98537 0.98574
2.2 0.98610 0.98645 0.98679 0.98713 0.98745 0.98778 0.98809 0.98840 0.98870 0.98899
2.3 0.98928 0.98956 0.98983 0.99010 0.99036 0.99061 0.99086 0.99111 0.99134 0.99158
2.4 0.99180 0.99202 0.99224 0.99245 0.99266 0.99286 0.99305 0.99324 0.99343 0.99361
2.5 0.99379 0.99396 0.99413 0.99430 0.99446 0.99461 0.99477 0.99492 0.99506 0.99520
2.6 0.99534 0.99547 0.99560 0.99573 0.99585 0.99598 0.99609 0.99621 0.99632 0.99643
2.7 0.99653 0.99664 0.99674 0.99683 0.99693 0.99702 0.99711 0.99720 0.99728 0.99736
2.8 0.99744 0.99752 0.99760 0.99767 0.99774 0.99781 0.99788 0.99795 0.99801 0.99807
2.9 0.99813 0.99819 0.99825 0.99831 0.99836 0.99841 0.99846 0.99851 0.99856 0.99861
3.0 0.99865 0.99869 0.99874 0.99878 0.99882 0.99886 0.99889 0.99893 0.99896 0.99900
3.1 0.99903 0.99906 0.99910 0.99913 0.99916 0.99918 0.99921 0.99924 0.99926 0.99929
3.2 0.99931 0.99934 0.99936 0.99938 0.99940 0.99942 0.99944 0.99946 0.99948 0.99950
3.3 0.99952 0.99953 0.99955 0.99957 0.99958 0.99960 0.99961 0.99962 0.99964 0.99965
3.4 0.99966 0.99968 0.99969 0.99970 0.99971 0.99972 0.99973 0.99974 0.99975 0.99976
3.5 0.99977 0.99978 0.99978 0.99979 0.99980 0.99981 0.99981 0.99982 0.99983 0.99983
3.6 0.99984 0.99985 0.99985 0.99986 0.99986 0.99987 0.99987 0.99988 0.99988 0.99989
3.7 0.99989 0.99990 0.99990 0.99990 0.99991 0.99991 0.99992 0.99992 0.99992 0.99992
3.8 0.99993 0.99993 0.99993 0.99994 0.99994 0.99994 0.99994 0.99995 0.99995 0.99995
3.9 0.99995 0.99995 0.99996 0.99996 0.99996 0.99996 0.99996 0.99996 0.99997 0.99997
Inverse lookup: P(Z<z) 0.800 0.850 0.900 0.950 0.975 0.990 0.995 0.999
z 0.84162 1.03643 1.28155 1.64485 1.95996 2.32635 2.57583 3.09023