CHOICE MODELS FOR PREDICTING DIVISIONAL WINNERS IN MAJOR-LEAGUE BASEBALL

Major league baseball in the United States is divided into two leagues and four divisions. Each team plays 16 games against teams in the sam e league. The winner in each division is the team winning the most gam es of the teams in that division. We wish to predict the division winn ers based on games played up to any specified time. We use a generaliz ed choice model for the probability of a team winning a particular gam e that allows for different strengths for each team, different home ad vantages, and strengths varying randomly with time. Future strengths a nd the outcomes of future games are simulated using Markov chain sampl ing. The probability of a particular team winning the division is then estimated by counting the proportion of simulated seasons in which it wins the most games. The method is applied to the 1991 National Leagu e season.