Game Theory for Managers

Topic: GAMES USING RANDOMISED STRATEGIES

Class: PGPM-TERM V
EXPECTED VALUE
Probability measures the likelihood of a particular event occurring. For instance, the
probability of drawing a spade from a deck of cards is 13/52 or 0.25. Since the outcome
of the event is uncertain, the payoffs are assessed using the concept of expected values.

E(x) = p1X1 + p2X2 +……+ pnXn

Consider a game of chance with 4 prizes ($1, $2, $3, and $4) where each outcome has a
specific probability of happening, shown in the table below:

Number Probability
$1 50%
$2 20%
$3 20%
$4 10%
 Pure Strategy: A strategy that a player plays with certainty.
Mixed Strategy: A player who randomizes her strategy.

Goalie
Dive Left Dive Right
Kick Left 0, 1 1, 0
Kicker
Kick Right 1, 0 0, 1

The winner of the penalty kick gets 1 and loser 0.

No Dominant strategy for kicker or goalie. No pure strategy Nash equilibrium
Table Matching cards
 Zero-sum Game and No Nash
equilibrium in pure strategies.
Player 2  The only way to play this game is to
Red Black play randomly.
 The probability of red or black is ½.
-1, 1 1, -1
 The Nash equilibrium is the
Red
probability for choosing between
Player 1

red and black that maximises the

1, -1 -1, 1 expected payoffs for both the
Black players i.e., (1/2, 1/2)
Table Rock-paper-scissors
Player 2
Rock Paper Scissors

0,0 -1, 1 1, -1
Rock
Player 1

1, -1 0, 0 -1, 1
Paper

-1, 1 1, -1 0, 0
Scissors
 Manager-worker Game
Table

Manager
Monitor Not Monitor

50, -10 50, 100

Work
Worker

-50, 10 100, -100

Shirk
Table
Manager-worker Game Manager’s expected payoffs from monitor:
-10p + 10(1-p)
Monitoring Game
Manager Manager’s expected payoffs from not monitor:
Monitor Not Prob. 100p -100(1-p)
Monitor If one of the payoffs is greater than the other, then it will be easy
for the manager to choose the strategy that gives a higher
50, -10 50, 100 p
Work expected payoff. Hence, the worker will mix the probability
Worker

between work and shirk in such a way that the expected payoff
-50, 10 100, -100 1-p from monitoring or not monitoring is the same for the manger.
Shirk
This implies that the expected payoffs are equated so that the
q 1-q manager is indifferent between the two strategies- monitor and
Prob. not monitor.
P= 0.5
Thus, the worker will work half the time and shirk half the time.

Worker’s expected payoffs from work:

50q + 50(1-q)
Worker’s expected payoffs from shirk:
-50q + 100(1-q)
q = 0.33

So, the worker will expect the manager to monitor with 1/3 probability and not to monitor with 2/3 probability.
The Nash equilibrium outcome is {p, 1-p} and {q, 1-q} i.e., {0.5, 0.5} and (0.33, 0.67}
 Random Deals
Table Random Deals Game

Reliance Retail

Deal No Deal Prob.

Deal 4, 3 5, 4 p
Big Bazar

3, 6 6, 5 1-p
No Deal

Prob. q 1-q
Big Bazar will equate its expected payoffs from deal and no deal to make it difficult for
reliance retail to predict about a deal:
4q + 5(1-q) = 3q + 6(1-q)
q = 0.5
Similarly, Reliance retail will equate its expected payoffs from deal and no deal to make
it difficult for Big Bazar to predict about a deal:
3p + 6(1-p) = 4p + 5(1-p)
p = 0.5
To see whether the two retailers are able to get better payoffs by randomizing their
strategies, we substitute the values of p and q in the expected payoff functions that
yield a payoff of 4.5 for both of them. Thus, a mixed strategy results in a worst
outcome.
Mixed and Pure Strategy Equilibria

Table This game has two Nash Equilibria, when

all choose Finance subject, and when all
Person 2 choose Economics. Thus, we have a
Finance Economics Coordination Problem. When will the two
students choose finance and when will
they choose Economics?
4, 3 1, 1
Finance
In order to find the Mixed Strategy
Person 1

Equilibrium, we first have to find the

0, 0 3, 4 probability that each of the players assigns
Econom
ics to each action. This will be done by
calculating the Expected Payoffs of Person
1, and Person 2 respectively.
 For Person 1, the Expected Payoff of this game is the
Table
sum of the payoffs of the two possible actions,
multiplied with the probability of Person 2 choosing
Person 2 those actions:
EP{Finance} = 4q + 1 (1 − q) = 3q + 1
Finance Economics
EP{Economics} = 0q + 3 (1 − q) = 3 − 3y
(q) (1-q)

4, 3 1, 1 3q + 1 = 3 − 3q
Finance
q=1/3
(p)
Person 1

Economic 0, 0 3, 4 We do the similar thing for Person 2, but this time we

s multiply his payoffs with the probabilities of Person 1:
(1-p) EP{Finance} = 3p + 0(1 − p) = 3p
EP{Economics} = 1p + 4(1 − p) = 4 – 3p
3p = 4 – 3p
P=2/3

For q = 1/3 and p = 2/3, the two students are indifferent between the two options. Therefore, those probabilities
are a Mixed Strategy Nash Equilibrium.
Think of a game being played between two app-preneurs who want to launch their apps.
Both apps are meant for book-lovers who can read books through exchanging books with
other users of the app, without having to become a member of a library. An app-preneur can
either go ahead and launch the app without fully testing all its features, or wait for the other
app-preneur to launch, and then, come out with a glitch-free app after observing and
rectifying the mistakes made by the rival.

Table App-preneurs Game

App-preneur 2

Launch Wait Prob.

Launch 0, 0 1, 2 p
preneur
App-

Wait 2, 1 -1, -1 1-p

q 1-q
The game has three Nash equilibria, two in pure strategies and one in mixed strategies. For the
values of p and q, where p = q = 0.5, we get the mixed strategy Nash equilibrium, where the two
best response curves for the players intersect. For p = 1 and q = 0, we get the pure strategy Nash
equilibrium {Launch, Wait}, and for p = 0 and q = 1, we get the pure strategy Nash equilibrium
{Wait, Launch}.

The game has two Nash equilibria in pure strategies.

0q + 1(1-q) = 2q – 1(1-q)
q = 0.5

0p + 1(1-p) = 2p – 1(1-p)
P = 0.5
 CASE STUDY 4.2
CHAMPIONS LEAGUE FINAL 2008
At the finals of the Champions League 2008, little known to the viewers, game theory was at work.
Before the match between Manchester United and Chelsea at Moscow’s Luzhinski Stadium, Chelsea’s
manager contacted economist Ignacio Palacios-Huerta, who had written a paper analysing penalty
shootouts. The economist had studied United’s team in detail and came up with four propositions: (i)
United’s goalkeeper, Edwin van der Sar mostly dived to the striker’s natural side (ii) Edwin van der Sar
stopped most of the strikes between 1 – 1.5 metres high (iii) United’s Cristiano Ronaldo hit 85 per cent of
the time to the right-hand corner if he stopped in the run-up to the ball (iv) The team that won the toss had
60 per cent chance of winning because of excessive pressure on the team going second.
The advice to Chelsea players was not to strike to their natural side, to hit low or high, but not in the
middle. United won the toss. The Chelsea players played according to the plan, except for Ashley Cole,
who hit to his natural side, but hit low so that it could not be grasped by van der Sar. Even Chelsea
captain, John Terry, who missed his strike, hit to the left. The penalty shootout went beyond the
mandatory five kicks, with both teams having missed one strike each. By the time Nicolas Anelka came,
the goalkeeper had realised five out of six strikes had been on the same side. It was then that he pointed to
Anelka that he knew he would place the ball in the left corner. Anelka panicked and hit the ball to the
right at the wrong height, missing the strike and disregarding the expert’s advice. Chelsea lost the match
and the title.
Mathematics of Mixed Strategies
Let us consider a constant-sum game related to terror attacks. Suppose a terrorist organization targets
army camps 90 per cent of the time and civilian areas 10 per cent of the time. Suppose also that
counter-terrorist commandos undertake combing operations in army bases 50 per cent of the time and
in civilian areas 50 per cent of the time. The payoff matrix for the terrorists in terms of success rates is
given as :

Table: Terrorist Attack Game

Terrorists

Army Civilian
50 100
Army
Commandos

90 80
Civilian
 Table: Terrorist Attack Game

Terrorists

Army Civilian p
50, 50 0, 100 0.5
Army
(0.45) (0.05)
Commandos

10, 90 20, 80 0.5

Civilian
(0.45) (0.05)
0.9 0.1
q

= 50 * 0.45 + 100 *0.05 + 90 * 0.45 + 80 * 0.05

= 22.5 +5 + 40.5 + 4 = 72
So, terrorist organization gets a payoff of 72 or the terrorist attacks are on average successful
72 per cent of the time. Because this is a constant sum, the expected payoff for counter-
terrorist commandos is 100 less the payoff for the terrorists i.e. 100-72 = 28.
What is the mixed strategy Nash equilibrium if the two online retailers want to
randomize their strategy?
Table

Snapdeal

Maximum Minimum
discount discount
Maximum -1, -1 3, 1
discount
Flipkart

1, 3 2, 2
Minimum
discount