Decision Modeling and Applications to Major League Baseball Pitcher Substitution

Relief pitcher substitution is an integral part of a Major League Baseball game as once a pitcher is changed, he cannot reenter the game. The decision to change a pitcher involves many factors and influences, including, but not limited to, the batter’s handedness, the state of the game, the pitcher’s tiredness, and the pitchers available in the bullpen. This research examines the influences that affect a manager’s decision to substitute a pitcher and presents an Analytic Hierarchy Process model to both incorporate these influences and analytically determine the best pitcher substitution, i.e., the best choice of pitcher to face a given batting line-up. The objective of this model as well as its extensions is to help managers make the best pitcher selections and staffing decisions to improve the probability of winning the game. Historical Major League Baseball box scores are used to both demonstrate the model and validate the manager’s decisions. This paper represents a somewhat unusual, although interesting and practical example of the use of decision modeling in a non-traditional application, namely Major League Baseball