Fictitious Play Outperforms Counterfactual Regret Minimization

Document pages: 6 pages

Abstract: We compare the performance of two popular algorithms, fictitious play andcounterfactual regret minimization, in approximating Nash equilibrium inmultiplayer games. Despite recent success of counterfactual regret minimizationin multiplayer poker and conjectures of its superiority, we show thatfictitious play leads to improved Nash equilibrium approximation over a varietyof game classes and sizes.