GAME THEORY
Dr Nabirasool
Assistant Professor
Indian Institute of Plantation Management,
Bengalore
Outline
What is Game Theory?
History of Game Theory and its Assumptions
Flow chart
Types of games/ Classification
Key Elements of a game theory
Significance & Limitations
Prisoners’ Dilemma
Nash equilibrium
Pure Strategies & Mixed Strategies
Methods of 2 players Zero-Sum games
Solutions of Pure Strategy Games
Principle of Dominance
Method of solving mixed strategy problems
What is game?
• Developed by Prof. John Von Neumann and Oscar
Morgenstern in 1928 game theory is a body of knowledge
that deals with making decisions when two or more rational
and intelligent opponents are involved under situations of
conflict and competition. The approach of game theory is to
seek to determine a rival’s most profitable counter- strategy to
one’s own best moves. It helps in determining the best course of
action for a firm in view of the expected counter moves from
the competitors.
History of game
theory
• von Neumann wrote a key paper in 1928
• 1944: “Theory of Games and Economic Behavior” by
von Neumann and Morgenstern
• 1950: Nash invents concept of Nash equilibrium
• Game theory booms after this…
• 1994: Harsanyi, Nash, and Selten win Nobel Prize in
economics for game theory work
ASSUMPTIONS
• There are finite number of competitors/ Participants
• There is conflict of interests between them.
• Each player has available with him finite courses of action/ choices
• Players know all possible available choices but does not know which one is
going to be chosen.
• Players simultaneously select their respective courses of action.
• The payoff is fixed and determined in advance/ gain or loss of each player is
finite and fixed before the game
• All players are equally wise
• Each player attempts to maximize his gains or minimize his loss
• Players have to make individual decisions without direct communication.
• A game involving ‘n’ players is called a n persons game
Decision
making
Under Uncertainity Under Risk Under Certainity
Non Competitive Competitive situations
Situations (Games Theory)
Pure Strategy Mixed
(Saddle Point exist)
Strategy
2*2 Strategies Game 2*n or 2*m strategies M*n strategies (Linear
(Arithmetic Method) game (Graphical Method) Programming Method)
Types of
game
• Two-Person Game – A game with 2 number of players.
• N-Person Game – A game with N number of players, where >2
• Cooperative: The players are assumed to be rational to realizes
that it is mutually advantageous to cooperate on any and every
action which is likely to benefit at least one of the players
without affecting the other adversely.
• Non-cooperative/ Prisoners’ Dilemma:
• 2-person Zero sum: A game in which sum of amounts won by
all winners is equal to sum of amounts lost by all losers.
Types of game
• Non-zero sum: A game in which the sum of gains and losses
is not equal.
• Pure-Strategy Game – A game in which the best strategy for
each player is to play one strategy throughout the game.
• Mixed-Strategy Game – A game in which each player
employs different strategies at different times in the game.
Key elements of a
game
• Game :
• Players: Who is interacting?/ A strategic decision maker within
the context of the game
• Interdependence:
• Strategies: What are their options?/ the course of action to be
taken by the oligopoly firms with the purpose of gaining most
from an action under the condition unknown reactions of the
rival firms.
• A complete plan of action a player/firm will take given the set of
circumstances that might arise within the game
• Pay off - It is the outcome of playing a game. It is the net gain, the
strategy brings to the firm for any given counter-strategy of the
competitor. The net gain is measured in terms of the objective of
the firm i.e., increase in profits, etc.
• Pay-off Matrix – It is the table showing outcomes or pay-offs of
different strategies of the game. E.g., pay-off matrix of a two-
person zero sum game
Player Y
Y1 Y2 Y3
Player X
X1 24 36 8
X2 32 20 16
In this pay-off matrix, positive pay-off is the gain to maximizing player (X)
and loss to minimizing player (Y).E.g., if X chooses strategy X1 and Y
chooses strategy Y1, then X’s gain is 24 and Y’s loss is 24.
• Information: What do they know?
• Rationality: How do they think?
• Equilibrium: the point in a game where both players have
made their decisions and an outcome is reached.
• Rule of dominance
Rule of dominance
•Rule of dominance: A strategy is called dominant if each pay-off in the
strategy is superior to each corresponding pay-off matrix, for player Y (who
is minimize) , strategy Y3 dominates both strategies Y1 and Y2. The rule of
dominance is used to reduce the size of pay-off matrix and thereby ease the
computational effort.
•If each element of a row (say I’th row) of the pay off matrix is less than
or equal to the corresponding element in the another row (average of the
corresponding elements of 2 or more rows then the I’th row) is said to be
dominated . Therefore I’th row can be deleted
•If each element of a Colum (say Jth colum) of the pay off matrix is
higher than or equal to the corresponding element in the another colum
(average of the corresponding elements of 2 or more colums then the Jth
colum) is said to be dominated . Therefore Jth colum can be deleted
Principle of Dominance
The Principle of Dominance states that if the strategy of a player
dominates over the other strategy in all condition, the later strategy can
be ignored because it will not affect the solution in any way. A strategy
dominates over other only if it is preferable over other in all conditions.
Following rules are used to reduce the size of pay-off:
Rule 1: If all the elements in a row of a pay-off matrix are lees than
or equal to the corresponding elements of other row then the player A
will never choose that strategy. So, former strategy is dominated by
latter strategy.
Rule 2: If all elements in a column in a pay-off matrix are greater than
or equal to the corresponding elements of other column then B will
never choose that column or former strategy is dominated by latter
strategy.
Rule 3: A Pure Strategy may be dominated if it is
inferior to average of two or more other pure strategies.
These rules could be better understood with the help
of the following example:
B1 B2 B3 B4
A1 7 6 8 9
A2 -4 -3 9 10
A3 10 2 -5 0
In this, B3 is dominating B4 which gives lesser loss in all
conditions. So, B4 is redundant ignoring this, the effective
pay-off will be:
B1 B2 B3
Row minima
A1 7 6 8 6
A2 -4 -3 9 -4
A3 10 5 -2 -2
Column 10 6 9
maxima
Now, the saddle point is where A’s optimum strategy is
A1 and of B is B2 and value of game is 6 on following the
rule of dominance. so, A cannot gain more than 6 and B
cannot loose less than 6.
Optimal strategy: A course of action or plan which puts the
player in the most preferred position irrespective if the strategy of
his competitors, is called an optimal strategy. Any deviation from
this strategy results in a decreased pay-off for the player.
Value of the game: it is the expected pay –off of the play when
all the players of the game follow their optimal strategies. The
game is called fair if the value of the game is zero and unfair if it is
non- zero.
Significance
Helps in decision making: Game theory develops a framework for
analysing decision makings under the situations of inter-
dependence of firms with existing uncertainties about the
competitor’s reactions to any course of action adopted by a firm.
Provide scientific quantitative technique: This theory outlines a
scientific quantitative technique which can be fruitfully used by
players to arrive at an optimal strategy, given firm’s objectives.
Gives insight into situation of conflicting interests: game theory
gives insight into several less-known aspects which arise in
situations of conflicting interests. For example, it describes and
explains the phenomena of bargaining and coalition-formation.
Limitations
• The assumption that the players have the knowledge about their own
pay-offs and pay- offs of others is rather unrealistic. He can only
make a guess of his own and his rivals’ strategies.
• As the number of maximum and minimax show that the gaming
strategies becomes increasingly complex and difficult. In practice,
there are many firms in an oligopoly situation and game theory
cannot be very helpful in such situation.
Limitations
• The assumptions of maximum and minimax show that the players
are risk-averse and have complete knowledge the strategies. These
do not seen practical.
• Rather than each player in an oligopoly situation working under
uncertain conditions, the players will allow each other to share the
secrets of business in order to work out a collusion. Thus, the
mixed strategies are also not very useful.
Pure
Strategies
In case of Pure Strategy game, maximizing player arrives at optimal strategy on the
basis of maximin criterion and minimizing player’s strategy is based on minimax
criterion. The game is solved when maximin value equals minimax value. This value is the
value of game. It could be explained using following example:
An
Example
Mixed
strategy
A mixed strategy game exists when there is no saddle point.
Each player will then optimize their expected gain by
determining the percent of time to use each strategy.
2-players Zero-Sum
games
Penny Matching:
• Each of the two players has a penny.
• Two players must simultaneously choose whether
to show the Head or the Tail.
• Both players know the following rules:
-If two pennies match (both heads or both
tails) then player 2 wins player 1’s penny.
-Otherwise, player 1 wins player 2’s penny.
Player 2
head tail
head -1,1 1,-1
Player 1
tail 1,-1 -1,1
Prisoner’s Dilemma
• No communication:
- Strategies must be undertaken without the full
knowledge of what the other players (prisoners)
will do.
• Players (prisoners) develop dominant strategies but are
not necessarily the best one.
Example
In order to understand prisoners’ dilemma let us suppose that there are 2
persons, A & B who are partners in an illegal activity of match fixing.
The CBI arrests them and lodges them in separate jails with no
possibility of communication between them. They are being interrogated
separated by CBI officials with. Following conditions are disclosed to
them:
1. If you confess your involvement in match fixing, you will get a 3 year
imprisonment.
2. If you deny your involvement and your partner denies too, you will be
set little punishment say 1 year each for lack of evidence.
3. If one of you confesses and turns approver, and other does not, then
one who confesses gets a free of walk as incentive, and the other one
gets 5 year imprisonment.
Both persons have 2 Actions/options: 1.To confess 2.Not to Confess
But both have a common objective to minimize the period of imprisonment.
Thus, following pay-off matrix is derived
B’soptions
Confess deny
Confess A B A B
A’s -3 -3 0 -4
options
Deny A B A B
-4 0 -1 -1
Thus, according to conditions there are many options available to them, one
being they both confess and get -3 year imprisonment as they do not know what
the other one will say. But the best they could achieve under given situations is,
they both deny and set -1 year in imprisonment. This situation as a whole is called
prisoners’ dilemma.
P2
Confess Deny
Confess -3,-3 0,-4
P1 Deny -4,0 -1,-1
Mathematical Notation:
Set of Players - { P1, P2}
Set of Rules
Set of Actions (A) : Ai denotes action set of player i
Here A1 (Action set of Player 1) {C, D} and A2 (Action set of player 2 ) {C,D}
Out comes is ‘O’, it is cartition product of both players actions that is , O=A1*A2
= { (C,C), (C,D), (D,C), (D,D)}
•What is Pay off the pay off can be represented as a function u i of the
out come
•Ui (o) , i.e U1(o), U2(o) these are the pay offs of different players Or Ui
(ai, a-i) it is the play off notation. U denotes Utility, ai denotes action of player
i, a-i denotes action of all players other than i.
U1 (c,c)=-3, U1(C,D)=0
U1(C,C)=
U1(C,D)=
U1(D,C)=
U1(D,D)=
Why is PD is in useful in practice
Example of Retail chain/store
R2
H L
H 500,500 0,750
R1
L 750,0 250,250
Nash’s Equilibrium
A Nash equilibrium is a situation in which none of them have
dominant Strategy and each player makes his or her best
response
(S, T) is Nash equilibrium if S is the best strategy to T and T
is the best strategy to S
John Nash shared the 1994 Nobel prize in
Economics for developing this idea!
This equilibrium occurs when each player’s strategy is
optimal, knowing the strategy's of the other players.
P2
Thought process of Player 1 Confess Deny
P2
Best Response or BR i.e., Confess -3,-3 0,-4
What is the BR Deny -4,0 -1,-1
BRi(a-i) BR of player 2
BRi is the best response of BR2(C)=
player i is the given the BR2(D)=
fixed action of all the other Best response dynamics
players (a-i) Nash Equilibrium (NE) is
BR of player 1 inter section of BR.
BR1(C)=C NE – each player is
BR1(D)=C playing his best response
to the action of all other
What is Nash Equilibrium?
Nash Equilibrium (NE) is inter section of BR.
NE – each player is playing his best response to the
action of all other players
NE has interesting implications
NE is SELF sustaining out come /enforcing
agreement
What do you mean by self enforcing agreement?
A sort of agreement which does not need any
other extraneous agency to enforce it, try to
understand better go back to PD game table
• Let us for instance assume that before going to interrogation rooms,
prisoner 1 and prisoner 2, agree to choose the out come deny, deny, which is
better for both of them because each one receives only a 1 year prison
sentence.
Is this agreement self enforcing?
• Well, not really
• To illustrate go back to PD Game table
• You can clearly see, no; because the moment they go into their interrogation
rooms prisoner 1 has an incentive to deviate; he was an incentive to deviate
to confess which will give him 0 years in prison; in fact prisoner 2 also has an
incentive to deviate which will change or reduce prison sentence from 1 year
to 0. So, even though they agree that they are going to choose D, this
agreement is not self sustaining because the moment they go into their
interrogation rooms they have an, each one has incentive to deviate.
• How about DC and CD
• Only CC agreement is self sustain, because no prisoner has an incentive to
deviate unilaterally.
What is another way to talk about a NE?
• A Nash equilibrium also as we said is that is where no player
has an incentive to deviate unilaterally, right.
• Unilaterally means by himself.
• Let us talk about that again.
Let us draw Game table
If you look at the out come CC prisoner 2 does not have an
incentive to deviate because if he deviates fro C to D he can get
-4, he is going to increase to his prison sentence to 4 years.
Similarly, if prisoner 1 deviates C to D he is going to increase his
prison sentence to -4. So, he does not have an incentive to
deviate.
How about DD?
How about CD?
How about DC?
• Another way to look at the Nash Equilibrium is that the Nash
equilibrium is a no regret.
• What do you mean by no regret outcome?
• Let us go and look at it again
• You can again see that the outcome is confess, confess then
none of prisoner1 or prisoner 2 have any regrets. Because you
cannot further improve your pay off.
• But if the out come is CD- what will happen
• Similarly if the outcome is DC- what will happen
• If outcome DD- what will happen, both
So, only C,C is only ‘No-regret’ outcome, which is the NE of this
game,
• Nash equilibrium:
• Action profile a1*, a2*……………………an* - NE
• Ui (ai* , a-i*) > Ui (ai , a-i*)
• If for each player i his pay of ui of ai star, a-i star,
remember this is the notation that we use, ai star
is action of player i an equilibrium action, a-istar is
the equilibrium action for all the other players, this
is greater than or equal to the payoff of player i
from any other action i give for a fixed action
profile a-istar of all the other players.
• This has to hold each player i and
• each action ai
Let us think about this
What does it mean?
It means that aistar that is given a-star, fixed a-star of all the other players, aistar
yields a higher payoff compared to any action, any other action ai for player i.
So, therefore,means aistar is the best response. This means, aistar is the best
response to, a-istar.
So, player i is playing his best response aistar.
And, this has to hold for all the players i which means each player has to play his
best response.
So, player I is playing the best response, and this is true for all the plyers i, that is
all players are playing their best response.
So, this is indeed the definition of the nash equilibrium.
So, the nash equilibrium can represented mathematically in the baove.
Coordination
games
coordination games are a class of games with multiple
pure strategy Nash equilibria in which players choose
the same or corresponding strategies.
situations in which all parties can realize mutual
gains, but only by making mutually consistent
decisions
Examples
B College MG road
Models the
A
strategic College 100,100 0,0
conflict when
two players
have to choose MG road 0,0 10,10
their
priorities
Pure coordination game
Exp2 –Stag hunt
A situation in which both Stag Hare
players (hunters) can
benefit if they cooperate Stag 10,10 0,7
(hunting a stag).
However, cooperation Hare 7,0 7,7
might fail, because each
hunter has an alternative
which is safer because it
does not require
cooperation to succeed
(hunting a hare).
Applications of Game
Theory
• Mathematics • Psychology
• Computer Science • Law
• Biology • Military Strategy
• Economics • Management
• Political Science • Sports
• International Relations • Game Playing
• Philosophy