Vous êtes sur la page 1sur 2

Modified Probability Tree for 8/3/18 Riddler Classic

I tried a few approaches before realizing that everything really hinges on getting tails, so I built a modified probability tree showing the combined ways to get the
next tail via the increasingly long heads streaks required. In this table,
 Every blue arrow is flipping TAILS; every red arrow flips HEADS. Every arrow has probability 1/2.
 The subscripts refer to the number of consecutive HEADS required to win the game relative to that current point in the game.
 The numbers above each Tn and below each H0 are the probabilities to get to that point.

1  1  1 1   1  3   1  3  1 1 1   1  3  7   1  3  7  15 
                         
2  2  2 4   2  4   2  4  2 4 8   2  4  8   2  4  8  16 
T2 T3 T4 T5
Start
H0
H1 H2 H3
(WIN)
1
 
2
H0 (WIN) H1 H2
2
 1  1 
  
 2  2 
H0 (WIN) H1
3
 1  3  1 
   
 2  4  2 
H0 (WIN)
4
 1  3  7  1 
    
 2  4  8  2 

So the probability of ending on a streak of heads one longer than the total number of tails flipped is the sum of all of the H0 probabilities:
2 3 4 5
 1   1  1   1  3  1   1  3  7  1   1  3  7  15  1 
                         ...
 2   2  2   2  4  2   2  4  8  2   2  4  8  16  2 

This series is painful, but the term pattern is clear. The halving factor at the end quickly decay and by 45 terms, my computer algebra software reaches its
computational limit or 0.71121190491339 .
Looking at this a different way, the only way to NOT win the game is to keep getting tails … and this becomes increasingly likely the longer the required heads
streak becomes. So, I could compute the probability of never getting a heads streak and calculate that complement. Turning to Wolfram Alpha, I get for the
product

And = 1-0.28878809508660242127889972… = 0.71121190291339757872110027… .

There might be a closed form for this, but I’m not seeing how to get a closed form from this. That said, I certainly see an elegant infinite product that gives the
answer.

I’d love to hear if anyone has a way to calculate an exact value….

Vous aimerez peut-être aussi