Three-Player Nash Equilibrium (Game Theory)

A three-player Nash equilibrium is a set of choices, one for each player, where no player can gain by changing their own choice alone while the other two keep theirs fixed. To find one, list each player’s actions and payoffs, identify mutual best responses, and, if needed, test mixed strategies using expected payoffs. Equilibrium does not mean the outcome is best for everyone.

Planning for future choices is easier when you know what an equilibrium can and cannot tell you. In economics, politics, and business, three parties may affect one another’s results. A sound analysis can show which choices are stable under the model. It cannot, by itself, prove what people will do or which outcome society should prefer.

I’ll use a small, fully specified game to show the method. The example is a model for learning, not a report about a real event. The same steps apply when you study a more complex game.

Three-player game setup and the equilibrium concept

A three-player strategic-form game names each player, lists the actions each can choose, and assigns a payoff to every resulting combination of actions. A strategy profile collects one choice per player. It is a Nash equilibrium when each choice is a best response to the other two choices.

Let the players be (i=1,2,3). Each has a strategy set (S_i), the list of actions available to that player. A profile is (s=(s_1,s_2,s_3)), and (s_{-i}) means the choices of everyone except player (i).

Each player has a payoff function (u_i(s_1,s_2,s_3)). It gives that player’s payoff for each possible profile. A payoff might represent money, points, or another measure of what the player values. The model should say what the numbers mean.

The defining test is:

[ u_i(s_i^,s_{-i}^) \geq u_i(s_i,s_{-i}^*) ]

for every player (i) and every alternative action (s_i) they could choose. In plain language, hold the other two players’ choices fixed. Ask whether this player could do better by switching alone. If no player can, the profile is a Nash equilibrium.

This is a test of unilateral deviation. It does not ask whether two or three players could improve their outcomes by changing together. That distinction matters: a profile can pass the Nash test and still leave room for a group to benefit from coordinating differently.

Before solving, state the game’s assumptions. For example, are choices made at the same time? Does each player know the available actions and payoff rules? Is the game one-shot or repeated? The answers affect how well the model fits a situation.

Payoff structure, best responses, and equilibrium conditions

A best response is a player’s most rewarding action given the other players’ choices. To find a pure-strategy equilibrium, compare each player’s payoffs across their available actions for every fixed pair of choices by the other players. A profile is an equilibrium only when all three chosen actions are best responses at once.

Consider a simple coordination game. Three players each choose A or B. Each earns 1 if all three choose the same action, and 0 otherwise. The payoffs below are listed in the order of Players 1, 2, and 3.

At (A, A, A), a player who switches alone creates a mismatch and gets 0 instead of 1. The same logic applies at (B, B, B). Both profiles are equilibria, so this game has more than one.

At (A, A, B), Player 3 can improve from 0 to 1 by switching to A. The profile fails the equilibrium test, even though Players 1 and 2 would not gain by switching alone. Every player must pass the test.

A useful way to work with a larger game is to mark best responses. For each fixed combination of two players’ actions, mark the action or actions that give the remaining player the highest payoff. A profile where all three players’ actions are marked is a pure-strategy Nash equilibrium.

Payoff numbers are usually cardinal when their size matters, as in expected-payoff calculations. They are ordinal when only the ranking of outcomes matters. Rankings are enough to identify best responses in a pure-strategy game. For mixed strategies, the numerical payoff differences matter, so the values must represent meaningful utility differences.

Solving the game: pure and mixed strategies

A pure strategy means choosing one action with certainty. A mixed strategy assigns probabilities to actions. To solve for a mixed-strategy equilibrium, calculate each player’s expected payoff and find probabilities that make every action used with positive probability equally attractive and no less attractive than an unused action.

In the coordination example, suppose each player independently chooses A with probability (p_i) and B with probability (1-p_i). If the other two players choose A with probabilities (p_j) and (p_k), Player (i)’s expected payoff from choosing A is (p_jp_k). The payoff from choosing B is ((1-p_j)(1-p_k)).

For a player to mix between A and B, these expected payoffs must be equal:

[ p_jp_k=(1-p_j)(1-p_k) ]

In a symmetric mixed equilibrium, all players use the same probability (p). The condition becomes (p^2=(1-p)^2), giving (p=0.5). Each player chooses A or B with equal probability. Against the other players’ equal mixes, neither action offers a higher expected payoff.

This example has two pure equilibria and a symmetric mixed equilibrium. A mixed equilibrium is not just a vague idea that players are “sometimes unpredictable.” It is a precise probability choice that makes each action used in the mix a best response.

A practical solving sequence is:

• List the players, their actions, the payoff for each profile, and assumptions about timing and information.
• Find each player’s best responses for every combination of the other two players’ actions.
• Identify profiles where all three choices are best responses.
• If no pure equilibrium appears, or if the question calls for mixed strategies, write expected payoffs and solve the indifference conditions.
• Check every candidate by substituting it into the payoff rules. Confirm that no player has a profitable unilateral deviation.

A finite game has at least one Nash equilibrium when mixed strategies are allowed. This result does not promise a pure equilibrium, and it does not promise a unique equilibrium. If calculations produce several candidates, test each one rather than choosing whichever looks most appealing.

Interpretation, applications, and limits of Nash equilibrium

A Nash equilibrium describes a stable pattern of choices under a specified model: no player can gain by changing alone. It is not a guarantee about what people will choose, and it does not mean the result is fair, efficient, or best for the group. Applications therefore require clear assumptions and careful interpretation.

Illustrative application: three firms choosing a standard. Suppose three firms must choose between two technical standards, and each benefits when all use the same one. This can be modeled like the coordination game. Both “all choose A” and “all choose B” may be equilibria. The model identifies coordination possibilities, but it does not explain which standard will win. History, communication, and switching costs would need to be included if they matter.

Illustrative application: three parties sharing a resource. A model might give each party a choice about how much to use and assign payoffs based on all three choices. The analyst can find stable choices, then compare them with a group objective such as total payoff. If a stable outcome uses too much of the resource, that would not contradict Nash equilibrium. Individual stability and group benefit are different questions.

These are stylized case studies, not claims about measured outcomes in actual industries or communities. Their purpose is to show how the method can organize a question. For a real case, the analyst must justify the action sets, payoffs, information assumptions, and timing.

Keep these limits in view:

• Equilibrium is not a prediction by itself. People may lack information, make mistakes, or respond to incentives missing from the model.
• Multiple equilibria leave a selection problem. The model may show several stable outcomes without saying which one will occur.
• Equilibrium is not social optimality. A profile can be stable against individual changes yet worse for everyone than another profile.
• Equilibrium is not a dominant-strategy result. A dominant strategy is best regardless of what others do; a Nash strategy need only be best given the particular choices or mixes in that equilibrium.
• Joint changes are outside the basic test. Nash equilibrium checks one player changing at a time, not a coalition changing together.

When interpreting an answer, I ask: What assumptions make this profile stable? Who could gain by changing alone? Are there other equilibria? Would the result change if information, timing, or payoffs changed? Those questions help prevent a correct calculation from being treated as a complete account of a real situation.

Frequently asked questions

These answers clarify common terms and checks used when solving three-player games. Each one focuses on what the equilibrium condition requires and what it does not establish.

What is a three-player Nash equilibrium?
It is a profile of choices where none of the three players can improve their payoff by changing their own choice alone, with the other two choices held fixed.

How do I test a pure-strategy equilibrium?
For each player, compare their payoff from the chosen action with the payoff from every available alternative, keeping the other two players’ actions fixed. The profile qualifies only if no player can gain by switching.

Can a three-player game have more than one equilibrium?
Yes. The coordination example has two pure equilibria and a mixed equilibrium. A game can also have no pure equilibrium, though every finite game has at least one equilibrium when mixed strategies are allowed.

Does Nash equilibrium mean everyone gets the best possible outcome?
No. It means no one can gain by changing alone. Another profile might give one or all players higher payoffs, but reaching it may require coordinated changes.

What is a mixed-strategy equilibrium?
It is a profile in which players use probabilities to choose among actions. Each action used with positive probability must provide the same highest expected payoff against the other players’ mixes.

Why do mixed strategies use expected payoffs?
A player compares the average payoff from an action across the possible actions of the other players. The probabilities and payoff values together determine that expected payoff.

Do payoffs have to be money?
No. They can represent any modeled measure of preference, such as points or utility. For mixed-strategy calculations, however, payoff differences must have a clear meaning, not just a ranking.

Can a Nash equilibrium predict what people will actually do?
Not on its own. It gives a result under stated assumptions. Real choices may depend on factors the model leaves out, such as limited information, changing incentives, or coordination.

(This article was written by one of our staff writers, Alan Westbrook. Visit our Meet the Team page.)