Mixed Strategies: Indifference Conditions 🎲

Introduction

students, imagine playing rock-paper-scissors against someone who keeps trying to predict your next move. If you always choose rock, they can beat you by choosing paper every time. So what if you sometimes choose rock, sometimes paper, and sometimes scissors in a carefully planned random way? That is the heart of mixed strategies in game theory.

In this lesson, you will learn how randomization can be a strategic choice, not just a lucky accident. The key idea is the indifference condition: a player mixes among strategies only when the other player’s choices make them equally happy with each option. If one option were clearly better, they would stop randomizing and choose the best one all the time.

Learning goals

By the end of this lesson, students, you should be able to:

• State the indifference condition for mixed equilibrium.
• Solve for equilibrium probabilities in simple games.
• Explain why indifference supports randomization.

What indifference means in a mixed strategy

A mixed strategy is when a player chooses between actions according to probabilities. For example, instead of always picking $A$, a player might choose $A$ with probability $p$ and $B$ with probability $1-p$.

The central idea behind a mixed equilibrium is this: if a player is mixing between two actions, then those actions must give the same expected payoff. Otherwise, the player would prefer the better one all the time. That is the indifference condition.

Think about buying snacks at school 🍎🍪. If one snack always tastes better and costs the same, you would pick that one every time. But if both options give you the same satisfaction, you might switch between them. In game theory, players randomize when the other side’s strategy makes them indifferent.

For a mixed equilibrium, each player’s randomization must make the opponent indifferent among the opponent’s own possible actions. This is the logic used to solve many simple games.

How to write the indifference condition

Suppose a player is deciding between two actions, $A$ and $B$. Let the opponent choose between two actions, $X$ and $Y$, with probabilities $q$ and $1-q$.

To find the value of $q$ that makes the player indifferent, we compare expected payoffs.

If the payoff from choosing $A$ is $u(A)$ and from choosing $B$ is $u(B)$, then the indifference condition is:

$$u(A)=u(B)$$

More specifically, if the payoffs depend on the opponent’s randomization, we compute the expected payoffs under $q$.

For example:

• Expected payoff from $A$: $E[u(A)] = q\cdot u(A,X) + (1-q)\cdot u(A,Y)$
• Expected payoff from $B$: $E[u(B)] = q\cdot u(B,X) + (1-q)\cdot u(B,Y)$

Then set them equal:

$$q\cdot u(A,X) + (1-q)\cdot u(A,Y)=q\cdot u(B,X) + (1-q)\cdot u(B,Y)$$

Solving this equation gives the mixing probability that keeps the player indifferent.

Example 1: A simple $2\times 2$ game

Let’s use a small payoff table. The first number in each cell is the payoff to the row player, and the second is the payoff to the column player.

| | Column $X$ | Column $Y$ |

|------------|------------|------------|

| Row $A$ | $(4,1)$ | $(1,3)$ |

| Row $B$ | $(2,2)$ | $(3,0)$ |

Suppose the column player chooses $X$ with probability $q$ and $Y$ with probability $1-q$.

To find the row player’s mixed strategy, we make the row player indifferent between $A$ and $B$.

Expected payoff from $A$:

$$E[u(A)] = 4q + 1(1-q)$$

Simplify:

$$E[u(A)] = 4q + 1 - q = 1 + 3q$$

Expected payoff from $B$:

$$E[u(B)] = 2q + 3(1-q)$$

Simplify:

$$E[u(B)] = 2q + 3 - 3q = 3 - q$$

Set them equal:

$$1+3q=3-q$$

Solve:

$$4q=2$$

$$q=\frac{1}{2}$$

So the column player must choose $X$ with probability $\frac{1}{2}$ and $Y$ with probability $\frac{1}{2}$ to make the row player indifferent.

That means the row player has no reason to prefer $A$ over $B$, because both give the same expected payoff.

Finding the other player’s probability

A mixed equilibrium usually requires solving for both players’ probabilities. Now let the row player choose $A$ with probability $p$ and $B$ with probability $1-p$.

To find $p$, we make the column player indifferent between $X$ and $Y$.

Expected payoff to the column player from choosing $X$:

$$E[v(X)] = 1p + 2(1-p)$$

Simplify:

$$E[v(X)] = p + 2 - 2p = 2 - p$$

Expected payoff from choosing $Y$:

$$E[v(Y)] = 3p + 0(1-p)$$

Simplify:

$$E[v(Y)] = 3p$$

Set them equal:

$$2-p=3p$$

Solve:

$$2=4p$$

$$p=\frac{1}{2}$$

So the mixed equilibrium is $p=\frac{1}{2}$ and $q=\frac{1}{2}$. Each player randomizes in a way that keeps the other player indifferent.

This is a powerful method: in many $2\times 2$ games, you solve one indifference equation for each player.

Why indifference supports randomization

Why would a player ever randomize if one action is better? The answer is simple: they usually would not. Randomization happens when the expected payoffs are exactly equal.

If a player mixing between two actions gets more payoff from one action, even by a little, then the player would shift all probability to that better action. So in equilibrium, the actions in the support of the mixed strategy must all give the same expected payoff.

This is why indifference is not just a math trick. It is the reason mixed strategies can exist at all. Randomization is stable only when no pure action in the mix is better than another.

A real-world example is guessing on a test when you are unsure between two answers. If both answers seem equally likely to be correct, you might choose one at random. In strategic settings, a player randomizes to avoid being predictable, but the randomization must be balanced so the opponent cannot exploit it.

A practical checklist for solving indifference problems

When students works on a mixed-equilibrium problem, this checklist helps:

1. Identify the strategies being mixed.

Usually these are two actions for each player.

1. Assign probabilities.

Let one player mix with probability $p$ and the other with probability $q$.

1. Write expected payoffs.

Compute the payoff from each action as a weighted average using the opponent’s probabilities.

1. Set the payoffs equal.

Use the indifference condition, such as $E[u(A)] = E[u(B)]$.

1. Solve for the probability.

Simplify the algebra carefully.

1. Check the result.

Make sure the probability is between $0$ and $1$.

If the solution gives a number outside the interval $[0,1]$, that usually means there is no interior mixed equilibrium in that game.

Example 2: A quick interpretation

Suppose a player is choosing between two actions, and the opponent mixes so that both choices give expected payoff $5$.

Then the player is indifferent because:

$$E[u(A)] = E[u(B)] = 5$$

This means the player can choose either action without changing the expected payoff. As a result, a mixed strategy can persist.

If instead one action gave expected payoff $6$ and the other gave $5$, then the player would choose the action with payoff $6$ every time, not mix. That would break the candidate equilibrium.

So the phrase “make the opponent indifferent” means “choose probabilities so the opponent cannot improve by switching to another pure action.”

Conclusion

Mixed strategies are a way to randomize strategically, not randomly for no reason. The core logic is the indifference condition: if a player is mixing, the expected payoffs of the actions they mix over must be equal. In simple games, this lets us solve for equilibrium probabilities by setting expected payoffs equal and solving the resulting equations.

students, whenever you see a mixed equilibrium problem, remember the main question: what probabilities make the other player just as happy with one action as with another? That is the key to finding equilibrium mixing probabilities and understanding why randomization can be rational 🎯

Study Notes

• A mixed strategy uses probabilities to choose between actions.
• The indifference condition says that actions in a player’s mix must give the same expected payoff.
• To find a mixed equilibrium, write the opponent’s expected payoff from each pure action.
• Set those expected payoffs equal, such as $E[u(A)] = E[u(B)]$, and solve for the probability.
• In a valid mixed equilibrium, the probabilities must lie in the interval $[0,1]$.
• If one action gives a higher expected payoff, the player will not keep mixing.
• Randomization is rational when it prevents the opponent from predicting and exploiting a pattern.
• Mixed equilibrium is supported by mutual indifference: each player’s mixing keeps the other player indifferent.