The existence of serial correlation and its effect on the format of a championship playoff series

The specific home/away sequencing of games has been a point of contention in the championship series of professional sports leagues. This research analyses data from the National Basketball Association (NBA) Finals in developing a logistic-regression model to predict the outcome of games that takes into consideration several factors, including home-court advantage and game-to-game momentum. It is found that negative serial correlation exists such that the team that wins one game is more likely to lose the next game of the series. This model is then used to predict the probability of each team winning each of the games within the series, and these probabilities are combined to evaluate alternative playoff formats. It is found that different formats are appropriate depending on the nature of momentum, and that the format used by the NBA may minimize travel requirements, but other formats may perform better in extending the length of the series.