A Bayesian analysis of the 2012 Pittsburgh Pirates.

摘要

This study examines statistics in Major League Baseball and their effects on team wins. Frequentist and Bayesian approaches are compared and past research is evaluated. Using Bayesian analysis and logistic regression, models are created for runs scored, runs allowed, wins and order of divisional finish. The models are subsequently used to analyze the performance of the Pittsburgh Pirates in 2011 and to predict their performance in 2012. The results reveal that the Pirates outperformed their expectations in 2011 based on their underlying statistics. Additionally, an initial model for wins indicates the Pirates will have their 20th consecutive losing season in 2012 based on their statistics, although a Bradley-Terry model suggests that their record should improve based as a result of playing weaker teams in the National League Central Division.