Fantasy baseball has been increasing in popularity dramatically over the past decade. The game begins with participants selecting a team through a fantasy draft at the beginning of the season and then tracking their players’ statistics, compared to those of their competitors, throughout the season. Selecting the players who will perform the best in the upcoming season is the goal at the start of the year, and there are many different rankings and algorithms devoted to assisting a participant in creating their team. This dissertation will focus on predicting the outcomes of the upcoming season and propose a selection algorithm based on the evaluations in order to optimize a participant’s fantasy draft. While the vast majority of fantasy baseball rankings do not disclose any analytical rigor behind them, this approach will focus on the methodology utilized to forecast player statistics throughout the upcoming season. A Bayesian approach will be utilized in conjunction with nonlinear growth curves and nonparametric regression tree approaches in order to predict future outcomes.
