Adaptation And Learning In Blotto Games

When Game Theory began to take off, the games that propelled it were The Prisoner’s Dilemma and Colonel Blotto. The Prisoner’s Dilemma, perhaps the most popular of games, thrived through much of its history, because of its broad applications. Its lesser known cousin, Blotto, sadly did not fare so well. In this paper, I show that equilibria can be found computationally by playing a large number of games. To do this, continuing with the military theme, I construct a model of a repeating Colonel Blotto game consisting of two players who are given troops to allocate across a given number of battlefronts based on predetermined strategies. In each game, the colonels add weight to the strategies they use and eventually they converge on optimal strategies. This research’s goal is to show that optimal strategies can be found computationally using a simple learning technique.