E-Nash equilibrium

From PokerAI

Jump to: navigation, search

Finding Nash equilibrium in almost any poker game played nowadays is computationally intractable. Therefore e-Nash equilibrium is often used for practical purposes. e is the value of the best response to the determined pseudo-optimal (or suboptimal) strategy and is a measure of how far from the actual equilibrium the strategy is.

The idea is that a particular strategy is an epsilon Nash equilibrium if it is known to have a minimum expectation within epsilon of the Nash equilibrium expectation. (Clearly, this only makes sense in games such as two-player zero sum games where all the Nash equilibria have the same expectation.) An advantage of using the epsilon-Nash equilibrium is that it can be used even in situations where the Nash expectation is unknown since it is often easier to determine a 'best response' than it is to find an equilibrium strategy.

See also