Machine learning implementations in baseball: an algorithmic prediction of the next pitch
University of New Brunswick
Machine Learning (ML) has recently begun gaining traction in the statistical analysis of baseball. Major League Baseball (MLB) has a long history of using statistics to evaluate players, but recent innovations in player tracking have introduced the opportunity for ML to flourish. Statcast is a new tracking system that generates detailed data pertaining to the movement of both the players and the ball using cameras and radar technology. This paper will examine the functionality of predictive models using this data and their applications in baseball. As an example, we will attempt to predict which type of pitch a pitcher will throw next. Random forest and support vector machine algorithms will be created for this learning task.