A mixed effects multinomial logistic-normal model for forecasting baseball performance

Abstract Prediction of player performance is a key component in the construction baseball team rosters. As result, most prediction models are proprietary property or industrial sports entities, and little known about them. Of those that have been published, main focus has to separately model each outcome with nearly no emphasis on uncertainty quantification. This research introduces joint modeling approach predict seasonal plate appearance vectors using mixed-effects multinomial logistic-normal model. accounts for positive negative correlations between outcomes, both across within seasons, provides posterior predictive distribution from which can be quantified. It applied important, yet unaddressed, problem predicting players moving Japanese (NPB) American (MLB) major leagues.