MODULE 5: GAME THEORY

Game theory
Game theory seeks to analyse competing situations which arise out of conflicts of interest. In
the present-day world, business organizations compete in getting the market share. The
conflicts of interests of human beings are not confined to the basic needs alone. Game theory
is another tool to examine situations of conflict to identify the courses of action to be
followed and to take appropriate decisions in the long run. Thus, this theory assumes
importance from managerial perspectives. This theory can offer valuable guidelines to a
manager in ‘strategic management’ which can be used in the decision-making process for
merger, take-over, joint venture, etc. The results obtained by the application of this theory
can serve as an early warning to the top level management in meeting the threats from the
competing business organizations and for the conversion of the internal weaknesses and
external threats into opportunities and strengths, thereby achieving the goal of maximization
of profits. While this theory does not describe any procedure to play a game, it will enable a
participant to select the appropriate strategies to be followed in the pursuit of his goals. The
situation of failure in a game would activate a participant in the analysis of the relevance of
the existing strategies and lead him to identify better, novel strategies for the future
occasions.
Game theory is generally defined as Type of decision theory in which one’s choice of action
is determined after taking into account all possible alternatives available to an opponent
playing the same game, rather than just by the possibilities of several outcomes.

Concepts in the Theory of Games:

➢ Players: The competitors or decision makers in a game are called the players of the
game.
➢ Strategies: The alternative courses of action available to a player are referred to as
his strategies.
➢ Payoff: The outcome of playing a game is called the payoff to the concerned
player.
➢ Optimal Strategy: A strategy by which a player can achieve the best payoff is
called the optimal strategy for him.
➢ Zero-sum game: A game in which the total payoffs to all the players at the end of
the game is zero is referred to as a zero-sum game.
➢ Non-zero-sum game: Games with “less than complete conflict of interest” are
called non-zero-sum games. The problems faced by many business organizations
come under this category. In such games, the gain of one player in terms of his
success need not be completely at the expense of the other player.
➢ Payoff matrix: The tabular display of the payoffs to players under various
alternatives is called the payoff matrix of the game.
➢ Pure strategy: If the game is such that each player can identify one and only one
strategy as the optimal strategy in each play of the game, then that strategy is
referred to as the best strategy for that player and the game is referred to as a game
of pure strategy or a pure game.
➢ Mixed strategy: If there is no one specific strategy as the ‘best strategy’ for any
player in a game, then the game is referred to as a game of mixed strategy or a
1
 mixed game. In such a game, each player must choose different alternative courses
of action from time to time.
➢ Saddle point: A saddle point of a game is that place in the payoff matrix where the
maximum of the row minima is equal to the minimum of the column maxima. The
payoff at the saddle point is called the value of the game and the corresponding
strategies are called the pure strategies.
a) A game is said to be fair if Maximum value = minimum value = 0
b) A game is said to be strictly determinable if Maximum value = minimum value
≠0
➢ Dominance: One of the strategies of either player may be inferior to at least one of
the remaining ones. The superior strategies are said to dominate the inferior ones.

Types of Games:
1. Two person games and n-person games: In two person games, there are exactly two
players and each competitor will have a finite number of strategies. If the number of
players in a game exceeds two, then we refer to the game as n-person game.
2. Zero sum game and non-zero-sum game: If the sum of the payments to all the
players in a game is zero for every possible outcome of the game, then we refer to the
game as a zero-sum game. If the sum of the payoffs from any play of the game is
either positive or negative but not zero, then the game is called a non-zero-sum game
3. Games of perfect information and games of imperfect information: A game of
perfect information is the one in which each player can find out the strategy that would
be followed by his opponent. On the other hand, a game of imperfect information is
the one in which no player can know in advance what strategy would be adopted by
the competitor and a player must proceed in his game with his guess works only.
4. Games with finite number of moves / players and games with unlimited number
of moves: A game with a finite number of moves is the one in which the number of
moves for each player is limited before the start of the play. On the other hand, if the
game can be continued over an extended period and the number of moves for any
player has no restriction, then we call it a game with unlimited number of moves.
5. Constant-sum games: If the sum of the game is not zero but the sum of the payoffs to
both players in each case is constant, then we call it a constant sum game. It is possible
to reduce such a game to a zero-sum game.
6. Non-constant games: Consider a game with two players. If the sum of the payoffs to
the two players is not constant in all the plays of the game, then we call it a non-
constant game.
7. Two-person zero sum game: A game with only two players, say player A and player
B, is called a two-person zero sum game if the gain of the player A is equal to the loss
of the player B, so that the total sum is zero.

Payoff matrix:
When players select their strategies, the payoffs (gains or losses) can be represented in the
form of a payoff matrix. If the game is zero sum, the gain of one player is equal to the loss of
other and vice-versa. Suppose A has m strategies and B has n strategies. Consider the
following payoff matrix. Player A wishes to gain as large a payoff aij as possible while player
2
B will do his best to reach as small a value aij as possible where the gain to player B and loss
to player A be (-aij).

Assumptions for two-person zero sum game:

For building any model, certain reasonable assumptions are quite necessary. Some
assumptions for building a model of two-person zero sum game are listed below.
a) Each player has available to him a finite number of possible courses of action.
Sometimes the set of courses of action may be the same for each player. Or, certain
courses of action may be available to both players while each player may have certain
specific courses of action which are not available to the other player.
b) Player A attempts to maximize gains to himself. Player B tries to minimize losses to
himself.
c) The decisions of both players are made individually prior to the play with no
communication between them.
d) The decisions are made and announced simultaneously so that neither player has an
advantage resulting from direct knowledge of the other player’s decision.
e) Both players know the possible payoffs of themselves and their opponents.

Minimax and Maximin Principle:

The objective of game theory is to know how these players must select their respective
strategies, so that they may optimize their payoffs. Such a criterion of decision making is
referred to as minimax-maximin principle. This principle in games of pure strategies leads to
the best possible selection of a strategy for both players. The Principle is as follows
• Consider player A & B
• Player A would like to maximize his minimum gain, the value is called maximin value
• Player B would like to minimize his losses, the value is called minimax value
• When these two are equal, the corresponding strategies are called optimal strategies &
game is said to have a saddle point
• The value of the game is given by saddle point (V).

3
Solution of games with Saddle point:

Example 1: Solve the game with the following pay-off matrix

Player B Strategies
1 2 3 4 5
Player A 1 -2 5 -3 6 7
Strategies 2 4 6 8 -1 6
3 8 2 3 5 4
4 15 14 18 12 20

Solution: Applying Minimax and Maximin Principle for the given Problem the condition is
Player A would like to maximize his minimum gain and Player B would like to minimize his
losses. Therefore, the matrix will be formed as follows:
First consider the minimum of each row and then find the maximum from it as shown

1 2 3 4 5 Maximin
1 -2 5 -3 6 7 -3
2 4 6 8 -1 6 -1
3 8 2 3 5 4 2
4 15 14 18 12 20 12

Next consider the maximum of each column and then find the minimum from it as shown

1 2 3 4 5 Maximin
1 -2 5 -3 6 7 -3
2 4 6 8 -1 6 -1
3 8 2 3 5 4 2
4 15 14 18 12 20 12
Minimax 15 14 18 12 20

We see that the maximum of row minima = the minimum of the column maxima. So the
game has a saddle point. The common value is 12. Therefore, the value V of the game = 12.
The best strategy for player A is strategy 4.
The best strategy for player B is strategy IV.
The game is favourable to player A.

Example 2: Player A can choose his strategies from (A1 A2 A3) only. While player B can
choose from the set (B1 B2) only. The rules of the game state that the payments should be
made in accordance with selection of strategies. What strategy should A & B play in order to
get the optimum benefits of the play.

4
 Strategy pair Payment to be made
selected
A1B1 Player A pays Rs.1 to player B
A1B2 Player B pays Rs.6 to player A
A2B1 Player B pays Rs.2 to player A
A2B2 Player B pays Rs.4 to player A
A3B1 Player A pays Rs.2 to player B
A3B2 Player A pays Rs.6 to player B

Solution: Since the problem is in the statement form, we have to form the payoff matrix
based on the condition that Player A wishes to gain large payoff as possible while player B
will do his best to reach smallest value as possible where the gain to player B and loss to
player A.
Player B
1 2
1 -1 6
Player A 2 2 4
3 -2 -6
Once the payoff matrix is formed, we apply Minimax and Maximin Principle to find the best
strategy
1 2 Maximin
1 -1 6 -1
2 2 4 2
3 -2 -6 -6
Minimax 2 6
Since Minimax = Maximin = 2, there exists a saddle point.
 value of the game is saddle point = 2 & the optimal strategy is (A2, B1)

Solution of Game without saddle point (Mixed strategy)

If the row minima is not equal to the column maxima then we say there is no saddle point,
the value of the game must be obtained by using the concept of mixed strategy. In this we
find the probabilities of the decisions the player might take and to apply probability the
matrix must be a square matrix of the order 2*2 as shown
Player B
B1 B2
Probability q 1-q
A1 p a b
Player A
A2 1-p c d

5
Let,
p = probability that player A choose the option 1
1-p = probability that player A choose the option 2
q = probability that player B choose the option 1
1-q = probability that player B choose the option 2

𝑑−𝑐
Probability 𝑝 =
(𝑎+𝑑)−(𝑏+𝑐)
𝑑−𝑏
Probability 𝑞 =
(𝑎+𝑑)−(𝑏+𝑐)
𝑎𝑑 − 𝑏𝑐
𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑔𝑎𝑚𝑒 𝑉 =
(𝑎 + 𝑑) − (𝑏 + 𝑐)
Example 3: Player A & B put a coin each simultaneously on the table without showing each
other. A wins Rs. 8 when both coins show head & Rs. When 1 when both coins are tails.
Player B wins Rs. 3 when the coins do not match. Prepare payoff matrix & determine best
strategies of each player & value of the game.
Player B
H T
H 8 -3
Player A
T -3 1

Solution: First we check if the problem has a saddle point or no (pure or mixed strategy).

Player B

H T Maximin
H 8 -3 -3
Player A
T -3 1 -3

Minimax 8 1

Since Minimax ≠ Maximin, there is no saddle point hence we must apply mixed strategy.
Therefore a = 8, b = -3, c = -3, d = 1, apply the probability formula we get

𝑑−𝑐
𝑝=
(𝑎 + 𝑑) − (𝑏 + 𝑐)
1 − (−3) 4
𝑝= =
(8 + 1) − (−3 − 3) 15
11
1−𝑝 =
15
𝑑−𝑏
𝑞=
(𝑎 + 𝑑) − (𝑏 + 𝑐)

6
 1 − (−3) 4
𝑞= =
(8 + 1) − (−3 − 3) 15
11
1−𝑞 =
15
𝑎𝑑 − 𝑏𝑐
𝑉=
(𝑎 + 𝑑) − (𝑏 + 𝑐)
8 × 1 − (−3 × −3) 1
𝑉= =−
(8 + 1) − (−3 − 3) 15

P = 4/15 suggests that the probability of player A choosing option 1 is 4 out of 15 times.
Q = 4/15 suggests that the probability of player B choosing option 1 is 4 out of 15 times.
1 – P = 11/15 suggests that the probability of player A choosing option 2 is 11 out of 15
times.
1 – Q = 11/15 suggests that the probability of player B choosing option 2 is 11 out of 15
times.
V = -1/15 suggests that Player A has a negative gain & loses on an average of -1/15 unit per
game. This is unfair to player A.

Principle of dominance to reduce the size of the game to 2*2 from a higher
order for a mixed strategy problem:
When dominance of a row (or a column) in the pay-off matrix occurs, we can delete a row
(or a column) from that matrix and arrive at a reduced matrix. This principle of dominance
can be used in the determination of the solution for a given game or reduce the higher order
matrix. The Rules of Dominance are as follows
✓ Rule 1: if each element in a row is less than or equal to the corresponding element in
another row, then that row is said to be dominated & can be removed from the matrix
✓ Rule 2: if each element in a column is greater than or equal to the corresponding
element in another column, then that column is said to be dominated & can be
removed from the matrix
✓ Rule 3: dominance need not be based on the superiority of pure strategy only. A given
strategy can be dominated if it is inferior to an average of two or more other pure
strategies.

Example 4: Use the dominance principle to solve the following game.

Player B
1 2 3 4 5
Player A 1 6 15 30 21 6
2 3 3 6 6 4
3 12 12 24 36 3

7
Solution: If we check for saddle point, we can see that it’s a pure strategy and if can be
solved directly, but since it is said use the principle we have to reduce the matrix to 2*2 and
then solve. Both cases the value will be the same.
Applying the dominance principle,
Since each element in the second row is less than or equal to the corresponding element in
the first row, the second row is said to be dominated & can be removed from the matrix.
1 2 3 4 5
1 6 15 30 21 6
3 12 12 24 36 3

Since each element in 2nd, 3rd and 4th column is greater than or equal to the corresponding
element in the 1st column, the 2nd, 3rd and 4th column is said to be dominated & can be
removed from the matrix
1 5
1 6 6
3 12 3

Now since the matrix is reduced to 2*2 we can check for pure or mixed strategy.
Player B

1 5 Maximin
1 6 6 6
Player A
3 12 3 3
Minimax 12 6

Since Minimax = Maximin = 6, there exists a saddle point.

 value of the game is saddle point = 6 & the optimal strategy is (A1, B5)

Example 5: Solve the following game, use the dominance method to reduce the matrix,
write the strategies adopted by each player & value of the game.
Player B
1 2 3 4
1 3 2 4 0
Player A
2 3 4 2 4
3 4 2 4 0
4 0 4 0 8
8
Solution: Applying Dominance Principle
Since each element in the 1st row is less than or equal to the corresponding element in the 4th
row, the 1st row is said to be dominated & can be removed from the matrix.

Player B
1 2 3 4
Player A 2 3 4 2 4
3 4 2 4 0
4 0 4 0 8

Since each element in 1st column is greater than or equal to the corresponding element in the
3rd column, the 1st column is said to be dominated & can be removed from the matrix

2 3 4
2 4 2 4
3 2 4 0
4 4 0 8
Adding the 3rd and 4th column elements and taking average

2 (3+4)/2
2 4 3
3 2 2
4 4 4
Since each element in 2nd column is greater than or equal to the corresponding to the
average of the elements in the 3rd and 4th column, the 2nd column is said to be dominated &
can be removed from the matrix

3 4
2 2 4
3 4 0
4 0 8
Adding the 3rd and 4th row elements and taking average we can see that each element in the
2nd row is equal to the average of the elements in the 3rd and 4th row. Hence, the 2nd row is
said to be dominated & can be removed from the matrix

3 4
3 4 0
4 0 8
Now since the matrix is reduced to 2*2 we can check for pure or mixed strategy.

9
 Player B

3 4 Maximin
3 4 0 0
Player A
4 0 8 0
Minimax 4 8

Since Minimax ≠ Maximin, there is no saddle point hence we must apply mixed strategy.
Therefore a = 4, b = 0, c = 0, d = 8, apply the probability formula we get

2
𝑝=
3
1
1−𝑝 =
3
2
𝑞=
3
1
1−𝑞 =
3
8
𝑉=
3
V = 8/3 suggests that Player A has a positive gain & gains on an average of 8/3 unit per
game. This is fair game to player A.

Graphical Method to reduce (2 *n) or (m * 2) matrix of the game to 2*2

matrix for a mixed strategy problem:
Example 6: Use graphical method to reduce the following game & hence solve.
Player B
1 2 3
Player A 1 3 -3 4
2 -1 1 -3

Solution: To solve this graphical we take probability that player A choose option 1 be x and
probability that player A chooses option 2 be 1-x, thus we can write the expected value of
player A when player B chooses his option as

10
B’s pure strategies A’s expected payoff E(x)

B1 E(x1) = 3x – 1(1-x) = 4x - 1
B2 E(x1) = - 3x + 1(1-x) = - 4x+ 1
B3 E(x1) = 4x – 3(1-x) = 7x + 3

The graph can be plotted as follows. Since the goal of Player A is to maximize the gain, we
have to go for maximin Criteria. Maximin criteria refers to lowest area and the highest point
in that lowest area. The maximum point is shown in the graph and the lines containing that
point reduces the matrix as follows.
1-x x
8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1

Maximum Point
0 0

-1 -1

-2 -2

-3 -3
Maximin

-4 -4

-5 -5

11
The extreme points of the lines are formed in the matrix as follows
Player B
2 3
Player A 1 -3 4
2 1 -3
Checking for pure or mixed strategy

Player B
2 3 Maximin
Player A 1 -3 4 -3
2 1 -3 -3
Minimax 1 4

Since Minimax ≠ Maximin, there is no saddle point hence we must apply mixed strategy.
Therefore a = -3, b = 4, c = 1, d = -3, apply the probability formula we get
4
𝑝=
11
7
1−𝑝 =
11

𝑞= 7
11
4
1−𝑞 =
11
−5
𝑉=
11
V = -5/11 suggests that Player A has a negative gain & loses on an average of -5/11 unit per
game. This is unfair game to player A.

Example 7: Obtain the optimal strategies for both persons & value of the game for two-
person zero-sum game, whose payoff matrix is as follows:
Player B
1 2
1 1 -3
2 3 5
Player A 3 -1 6
4 4 1
5 2 2
6 -5 0
12
Solution: To solve this graphical we take probability that player B choose option 1 be y and
probability that player B chooses option 2 be 1-y, thus we can write the expected value of
player B when player A chooses his option as

A’s pure strategies B’s expected payoff E(y)

A1 E(y) = y – 3(1-y) = 4y - 3
A2 E(y) = 3y + 5(1-y) = - 2y + 5
A3 E(y) = -y + 6(1-y) = -7y + 6
A4 E(y) = 4y + (1-y) = -3y + 1
A5 E(y) = 2y + 2(1-y) = 2
A6 E(y) = -5y + 0(1-y) = -5y

1-y y
8 8

7 7
Minimax
6 6
Minimum Point

5 5

4 4

3 3

2 2

1 1

0 0

-1 -1

-2 -2

-3 -3

-4 -4

-5 13 -5
Since the goal of Player B is to minimize the gain, we must go for minimax Criteria.
Minimax criteria refers to topmost area and the lowest point in that lowest area. The
minimum point is shown in the graph and the lines containing that point reduces the matrix
as follows.
Player B
1 2
Player A 2 3 5
4 4 1
Checking for pure or mixed strategy

Player B
1 2 Maximin
Player A 2 3 5 3
4 4 1 1
Minimax 4 5

Since Minimax ≠ Maximin, there is no saddle point hence we must apply mixed strategy.
Therefore a = 3, b = 5, c = 4, d = 1, apply the probability formula we get

V = -suggests that Player A has a negative gain & loses on an average of unit per game. This
is unfair game to player A.
p = 3/5
1-p = 2/5
q = 4/5
1-q = 1/5
V = 17/5
V = 17/5 suggests that Player A has a positive gain & gains on an average of 17/5 unit per
game. This is fair game to player A.

14