In many strategic situations, being predictable is a losing strategy. If your opponent knows exactly what you're going to do, they can easily counter it. This is where the idea of a mixed strategy comes in. Instead of choosing one specific action (a pure strategy), a player chooses a probability distribution over their available actions. They decide to play each of their pure strategies with a certain probability, effectively randomizing their choice.

Consider the simple game of Matching Pennies. Two players, Player 1 and Player 2, each secretly place a penny down, either heads up or tails up. If the pennies match (both heads or both tails), Player 1 wins Player 2's penny. If they don't match, Player 2 wins Player 1's penny. This is a game of pure conflict. If you look for a pure strategy Nash equilibrium, you won't find one. If Player 1 always chooses Heads, Player 2's best response is to always choose Heads. But if Player 2 always chooses Heads, Player 1's best response is to switch to Tails. There's no stable outcome where both players are happy with their single choice.

The only way to find a stable equilibrium is for both players to play randomly, choosing Heads 50% of the time and Tails 50% of the time. When they do this, neither player can improve their expected payoff by changing their own randomization. This stable point is called a Mixed Strategy Nash Equilibrium (MSNE).

How do we find the exact probabilities for an MSNE? The key is the Indifference Principle A player will only be willing to randomize between several pure strategies if they get the same expected payoff from each of them, given the other player's strategy. If one strategy gave a higher expected payoff, the player would simply choose that strategy all the time, and we'd be back to a pure strategy.

Let's use this principle to calculate the MSNE for Matching Pennies. Let's say Player 2 plays Heads with probability $q$ and Tails with probability $1-q$. Player 1 needs to be indifferent between playing Heads and playing Tails. We can set up equations for Player 1's expected payoff ($E_1$) for each of their pure strategies.

For Player 1 to be indifferent, these expected payoffs must be equal.

Mixed strategies aren't just for conflict games. They also appear in coordination games like the Battle of the Sexes. In this game, two people want to meet up, but have different preferences for the location. Let's say one prefers the Opera and the other prefers the Football game. They both get a high payoff if they go to the same place, but a higher payoff if it's their preferred place. If they go to different places, they both get zero.

This game has two pure strategy Nash equilibria: (Opera, Opera) and (Football, Football). But there is also a third, mixed strategy equilibrium. Let's find it. Assume Partner 1 goes to the Opera with probability $p$ and Partner 2 goes to the Opera with probability $q$.

To make Partner 2 indifferent, we set their expected payoffs from choosing Opera and Football equal, given Partner 1's strategy $p$.

Now, to make Partner 1 indifferent, we set their expected payoffs from choosing Opera and Football equal, given Partner 2's strategy $q$.

So, the Mixed Strategy Nash Equilibrium is (Partner 1 plays Opera with $p=3/5$, Partner 2 plays Opera with $q=2/5$). In this equilibrium, both players are randomizing in a way that keeps the other indifferent. However, this equilibrium is inefficient. The probability that they miscoordinate and both get zero is $p(1-q) + (1-p)q = (3/5)(3/5) + (2/5)(2/5) = 13/25$, which is over half the time! This is why, in real life, people try to communicate to land on one of the better pure strategy equilibria.

Let's test your understanding of these concepts.

Understanding mixed strategies adds a crucial layer to game theory. It allows us to find stable solutions in games where predictable, pure strategies would lead to an endless cycle of players trying to outguess each other. By embracing calculated randomness, players can achieve a strategic balance that is otherwise impossible.