copyright robert j. marks ii ece 5345 multiple random variables: a gambling sequence
TRANSCRIPT
copyright Robert J. Marks II
ECE 5345Multiple Random Variables: A Gambling Sequence
copyright Robert J. Marks II
Optimal % Betting
• You play a sequence of Bernoulli games with the chance of you winning = p >1/2. (Why must his be?)
• You start with a stash of $D. • If you bet and loose, you loose what you
bet. If you win, your bet is matched.• What is the optimal % bet for steady state
winning?
copyright Robert J. Marks II
Optimal % Betting
• Example Sequence– Win D[1]= D + %D =(1+%)D– Loose D[2]=D[1]- %D[1]= (1+%) (1-%)D– Loose D[3]=D[2]- % D[2]= (1+%) (1-%)2D – Win D[4]= (1+%) 2 (1-%)2D
• n trials and k wins leaves a stash ofD[n]=(1+%) k (1-%)n-k D
copyright Robert J. Marks II
Optimal % Bettingn trials and k wins leaves a stash of
D[n]=(1+%) k (1-%)n-k D
What value of % maximizes D[n] for large n?
Note: Maximizing D[n] for a fixed n is the same as maximizing
)Dlog(%1log%1log
][Dlog1
][
n
kn
n
k
nn
nB
copyright Robert J. Marks II
Optimal % Betting
The law of large numbers says
and
Thus
)Dlog(%1log)1(%1log][ ppnB
pn
k
k
lim p
n
kn
k
1lim
%1
1
%10
%
][
pp
d
ndB
copyright Robert J. Marks II
Optimal % Betting
Solving
The optimal return is then
D[n] = (1+%) k (1-%)n-k D = 2n pk (1-p)n-k D
Since
12% p
1/n1/1 D)1(2][D ppn ppn
Sanity Check:
100% for p=1
0% for p=1/2
copyright Robert J. Marks II
Optimal % Betting
The number of turns to double your stash is obtained by setting this to two and solving for n. The result is
npp ppn )1()1(2
D
][D
122double )1(log)1(log1 ppppn
copyright Robert J. Marks II
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 10
5
10
15
20
25
30
35
Optimal % Betting 1
22double )1(log)1(log1 ppppn
p
Notes:
At p=1/2, ndouble =
At p=1, ndouble = 1
copyright Robert J. Marks II
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 110
0
101
102
103
104
Optimal % Betting
p
122double )1(log)1(log1 ppppn
copyright Robert J. Marks II
Roulette
p = even or green = 20/36 = 0.5557
% = 2p-1 = 11.13%
ndouble = 111.655
copyright Robert J. Marks II
What if there are odds?
Extend the problem to where you loose % of your bet when you loose and get w% of your bet when you win. Either or w can exceed one.
What is the relation among w, ,and p to assure you are in a position to win.
Compute the optimal % of your bet.
Optimal % Betting