An Extensive Investigation of Strategies in Baseball

Research Paper will be posted in the coming weeks. Check back soon!
Download the
Full Paper Here
Authors

Melville, William; Mott, Tristan; Grimsman, David; Archibald, Christopher

Abstract

We model baseball as a zero-sum, extensive-form game played between two managers who each aim to play the strategy that leads their team to victory. Unfortunately, like chess and other large extensive-form games, baseball’s state space is too large to solve for a true optimal strategy in a reasonable amount of time. Thus, we implement five game-playing algorithms to efficiently approximate the optimal strategy. We call these algorithms our AI managers, and we evaluate them through thousands of simulated games. We demonstrate the usefulness of the winning AI manager by reevaluating Kevin Cash’s controversial decision to relieve Blake Snell in the 2020 World Series and Aaron Boone’s controversial use of Nestor Cortes in the 2024 World Series.