Professional Documents
Culture Documents
Game Theory 2
1
Quotes of the Day
New Quotes
2
From Marilyn Vos Savants column.
Say youre in a public library, and a beautiful stranger strikes
up a conversation with you. She says: Lets show pennies to
each other, either heads or tails. If we both show heads, I pay
you $3. If we both show tails, I pay you $1. If they dont match,
you pay me $2.
At this point, she is shushed. You think: With both heads 1/4
of the time, I get $3. And with both tails 1/4 of the time, I get $1.
So 1/2 of the time, I get $4. And with no matches 1/2 of the
time, she gets $4. So its a fair game. As the game is quiet,
you can play in the library.
Marilyn Vos Savant has a weekly column in Parade. She has the
highest recorded IQ on record.
Parade Publications, Inc. All rights reserved. This content is excluded from our Creative
Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. 3
Payoff (Reward) Matrix for Vos Savants Game
Beautiful Stranger
C1 C2
Heads Tails
This matrix is the
3 -2 payoff matrix for you,
R1: Heads and the beautiful
stranger gets the
R2: Tails -2 1 negative.
4
The Linear Program
C1 C2
3 -2 p
-2 1 1-p
0p1
5
Key Observation
When there are only two rows, the only variables for the
LP are z and p.
One can create a two dimensional drawing of the LP.
There is an equivalent but more standard approach.
6
Determining the optimal strategy
B. S.
H T Prob
Choose the value of p that
H 3 -2 p maximizes the minimum payoff.
T -2 1 1-p
Col 1 Col 2
maximize z = min {3p + -2(1-p), -2p + 1(1-p) }
subject to 0p 1
Col 1 Col 2
maximize z = min {5p 2, -3p + 1 }
subject to 0p 1 7
Determining the optimal strategy
Col 1 Col 2
maximize z = min {5p 2, -3p + 1 }
subject to 0p 1
3 Column 1 5p 2
payoff to you.
2
1
0
Column 2
-1
-2 -3p + 1
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
p 8
The beautiful strangers viewpoint
B. S.
H T
Choose the value of y that
H 3 -2 minimizes that maximum payoff.
T -2 1
Prob y 1-y
Row 1 Row 2
Row 1 Row 2
minimize z = max {5y 2, -3y + 1 }
subject to 0y 1 9
The Beautiful Strangers Viewpoint
Row 1 Row 2
minimize w = max {5y 2, -3y + 1 }
subject to 0y 1
3 Row 1 5y 2
payoff to you.
2
1
0
Row 2
-1
-2 -3y + 1
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
y 10
The payoffs are the same when y = 3/8
3 Row 1 5y 2
payoff to you.
2
1
0
Row 2
-1
-2 -3y + 1
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
y 11
A difficulty with mixed strategies in practice
\ \
13
Max consecutive string of \ or out of 20
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15
13 19 23
A game involving bluffing
(and asymmetric information)
Next: an example based on bluffing in poker.
No double
$100
Double accepted
-$200
You double
Coin is tails
Not accepted
$100
No double
-$100
Is bluffing a good idea?
A B
Heads Double the bet
.5
$200 with H or T $0 $100
A
Tails C D
.5
-$200 Double with H,
not with T $50 $0
C D
Let y be the probability of $50 $0
doubling when the coin is a tails.
Doubles
$100 are not
accepted.
payoff to you.
$50
Doubles
are
accepted.
$0
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
y 24
How frequently should your
opponent accept doubles? A B
double
$0
Let w be the probability with H or T $100
of accepting a double, if double C D
it is offered. with H $50 $0
$100
payoff to you.
You double
with H.
$50
You double
with H or T.
$0
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
w 25
A comment on bluffing
1. Highway
2. Local roads
3. MBTA
Robust optimization with mixed strategies
Good day Bad for Bad for local Bad for
highway roads MBTA
Highway 30 80 30 30
Local roads 40 40 90 40
MBTA 50 50 50 75
1. Highway
2. Local roads
3. MBTA
4. All probabilities are 1/3.
An optimal mixed strategy
Prob Good day Bad for Bad for Bad for
highway local roads MBTA
Highway 1/4 30 80 30 30
Local roads 1/4 40 40 90 40
MBTA 1/2 50 50 50 75
Average 42.5 55 55 55
Applications to games
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.