Analyzing mixed strategy games requires venturing beyond pure strategies and delving into probabilistic approaches. Here’s an overview of some common methods for solving mixed strategy games, excluding saddle point methods which focus on pure strategies:
1. Dominance (Algebraic Method):
This method identifies and eliminates strategies that are strictly dominated by others for a particular player. Here’s the approach:
- Analyze the payoff matrix row-wise for the maximizing player (Player A) and column-wise for the minimizing player (Player B).
- Identify any row (for Player A) where there exists another row with consistently higher payoffs in every column. This dominated row can be eliminated as Player A wouldn’t choose it.
- Similarly, identify any column (for Player B) where another column offers consistently lower payoffs across all rows. This dominated column can be eliminated as Player B wouldn’t choose it.
- Repeat the process of checking for dominance after eliminating dominated rows/columns.
- Once no more dominated strategies are found, set up a system of linear equations based on the remaining strategies and the equilibrium conditions (expected payoff for each player must be equal across all their strategies).
- Solve the system of equations to determine the optimal mixed strategies (probabilities assigned to each remaining strategy) for both players and the value of the game (expected payoff for each player at equilibrium).
2. Graphical Method (for 2xN or Mx2 Games):
This method is applicable for specific game sizes (two rows for Player A and N columns or M rows for Player A and two columns for Player B). Here’s the process:
- Plot the average payoff for Player A (considering all possible mixed strategies of Player B) for each of their pure strategies on a graph.
- The optimal mixed strategy for Player A corresponds to the point where the average payoff curve intersects a horizontal line representing the value of the game (which can be initially estimated).
- Once Player A’s strategy is determined, a similar graphical analysis can be done for Player B to find their optimal mixed strategy.
3. Odds Method (for 2×2 Games):
This method is specifically designed for 2×2 games. Here’s how it works:
- Calculate the difference between the highest and lowest payoffs in each row (for Player A) and each column (for Player B).
- Divide the difference in each row (Player A) by the sum of all the differences in that row. This gives the initial probabilities for Player A’s pure strategies.
- Similarly, divide the difference in each column (Player B) by the sum of all the differences in that column. This provides the initial probabilities for Player B’s pure strategies.
- Check if the expected payoffs for both players (considering the assigned probabilities) are equal. If not, adjust the probabilities proportionally until the expected payoffs converge for both players.
Important Notes:
- These methods become more complex for larger games (MxN) and might require advanced techniques like linear programming or iterative methods.
- The dominance method offers a clear algebraic approach but can be tedious for larger games.
- The graphical method provides a visual understanding but is limited to specific game sizes.
- The odds method is efficient for 2×2 games but not directly applicable to larger games.
Remember, the choice of method depends on the specific game size and complexity. For a well-rounded understanding, consider exploring resources that demonstrate the application of each method with example problems.