Home advantage in elimination games and the NBA play-in tournament

Abstract

Some of the most exciting contests in US professional basketball are the win-or-gohome, game seven’s in the playoffs. The National Basketball Association has instituted a play-in tournament for the 2021 and 2022 playoffs consisting solely of one-game series, which has sparked considerable controversy among the league’s executives and players. To understand the effect of the play-in tournament on playoff participation, we developed a model to determine the home-court advantage in elimination games. Various solution techniques—including the log-binomial and robust Poisson regression models—are used to estimate the model parameters using elimination-game data from the 1955–2019 playoffs. These models are appropriate alternatives to logistic regression as probabilistic classifiers with dichotomous response variables and provide risk ratios (in terms of probabilities) that are easier to interpret for someone unfamiliar with odds ratios. Results indicate that the home-court advantage for equally matched teams would be in the 0.50 to 0.55 range; when considering the games are played at the home arena of the team with the better regular-season record, the home team is expected to win 65 per cent of elimination playoff games. These models can be used to estimate the likelihood of each play-in participant making the playoffs.