Mathwords logoMathwords

Bimatrix Game — Definition, Formula & Example

A bimatrix game is a two-player game where each player has a finite list of strategies and has their own payoff table. One matrix holds the row player's payoffs and a second matrix holds the column player's payoffs, so the two players' interests do not have to be exact opposites.

A bimatrix game is a finite two-player game in normal form specified by a pair (A,B)(A, B) of real m×nm \times n matrices. Player 1 chooses a row i∈{1,…,m}i \in \{1,\dots,m\} and Player 2 chooses a column j∈{1,…,n}j \in \{1,\dots,n\}. Player 1 then receives aija_{ij} and Player 2 receives bijb_{ij}. A pair of mixed strategies (x∗,y∗)(\mathbf{x}^*, \mathbf{y}^*) is a Nash equilibrium if x∗TA y∗≥xTA y∗\mathbf{x}^{*T} A\,\mathbf{y}^* \ge \mathbf{x}^{T} A\,\mathbf{y}^* for every mixed strategy x\mathbf{x} and x∗TB y∗≥x∗TB y\mathbf{x}^{*T} B\,\mathbf{y}^* \ge \mathbf{x}^{*T} B\,\mathbf{y} for every mixed strategy y\mathbf{y}. The special case B=−AB = -A is a zero-sum (matrix) game.

Key Formula

u1(x,y)=xTA y,u2(x,y)=xTB yu_1(\mathbf{x},\mathbf{y}) = \mathbf{x}^{T} A\,\mathbf{y}, \qquad u_2(\mathbf{x},\mathbf{y}) = \mathbf{x}^{T} B\,\mathbf{y}
Where:
  • x\mathbf{x} = Row player's mixed strategy: a column vector of probabilities over the m rows that sums to 1
  • y\mathbf{y} = Column player's mixed strategy: a column vector of probabilities over the n columns that sums to 1
  • AA = Row player's m × n payoff matrix
  • BB = Column player's m × n payoff matrix
  • u1,u2u_1, u_2 = Expected payoffs to the row player and the column player

Worked Example

Problem: Two players each pick Top/Bottom (row player) or Left/Right (column player). The payoff matrices are A=(3002)A = \begin{pmatrix} 3 & 0 \\ 0 & 2 \end{pmatrix} for the row player and B=(2003)B = \begin{pmatrix} 2 & 0 \\ 0 & 3 \end{pmatrix} for the column player. Find all Nash equilibria.
Step 1: Check pure strategies. At (Top, Left) the payoffs are (3, 2). Switching drops the row player to 0 or the column player to 0, so neither player wants to switch alone. The same reasoning works at (Bottom, Right) with payoffs (2, 3). The two off-diagonal cells pay (0, 0), and either player can improve by switching. So the game has two pure equilibria.
(Top,Left)→(3,2),(Bottom,Right)→(2,3)(\text{Top},\text{Left}) \to (3,2), \qquad (\text{Bottom},\text{Right}) \to (2,3)
Step 2: Look for a mixed equilibrium. Let the row player play Top with probability pp. In equilibrium, pp must make the column player indifferent between Left and Right. That condition uses matrix BB.
2p=3(1−p)  ⇒  p=352p = 3(1-p) \;\Rightarrow\; p = \tfrac{3}{5}
Step 3: Let the column player play Left with probability qq. This qq must make the row player indifferent between Top and Bottom, so the condition uses matrix AA.
3q=2(1−q)  ⇒  q=253q = 2(1-q) \;\Rightarrow\; q = \tfrac{2}{5}
Answer: The game has three Nash equilibria: (Top, Left), (Bottom, Right), and the mixed equilibrium where the row player plays Top with probability 3/5 and the column player plays Left with probability 2/5. In the mixed equilibrium, each player's expected payoff is 65\tfrac{6}{5}.

Why It Matters

Bimatrix games model situations where the players' goals partly align and partly conflict. Examples include firms choosing technology standards and countries negotiating trade policy. These cases fall outside zero-sum games. Computing their equilibria is a common topic in game theory courses; methods such as the Lemke–Howson algorithm appear in algorithmic game theory and operations research.

Common Mistakes

Mistake: Treating a bimatrix game like a zero-sum game: assuming B=−AB = -A, or looking for a single value of the game.
Correction: The column player's matrix BB is its own table; only a zero-sum game has B=−AB = -A. A bimatrix game can have several Nash equilibria with different payoffs. Here they pay (3,2)(3, 2), (2,3)(2, 3) and (65,65)\left(\tfrac{6}{5}, \tfrac{6}{5}\right), so there is no single value of the game.

Related Terms

  • Game Theory — Parent field that studies bimatrix games and equilibria
  • Prisoner's Dilemma — Famous 2×2 bimatrix game with a dominant strategy
  • Pareto Optimality — Outcome where no one can be made better off without making someone worse off
  • Payoff — Number each player receives for a strategy outcome
  • Mixed Strategy — Randomized play, as in this game's third equilibrium