Gentleman's Sweep: Empirical Evidence from the National Basketball Association

Abstract

Under the current 2:2:1:1:1 NBA hosting arrangement, a team that is leading 3-0 has financial incentive not to win Game Four and to prolong the series to host Game Five at home. The incentive to prolong the series has to be balanced against the risk of losing the series and the potential impact of a prolonged series on subsequent series should the team progress. Extending the series to host the last and final game at home is referred to the Gentleman's sweep.In this work, we empirically investigate the gentleman's sweep in the NBA in the period 2003 to 2013. We show that the outcome of the third game is a strong indicator of whether the series extends to more than four games. Since Games Three and Four are played in similar conditions, a close game three indicates that the two teams are evenly matched and that game four could go either way. A blow out in Game Three indicates that the leading team has a significant advantage and is favored to win Game Four. In the last ten years, only three leading teams lost Game Four after convincingly winning Game Three. In same period, 19 leading teams won Game Four after convincingly winning Game Three.We then investigate the impact of Game Four on the current and future series. Our results are consistent with prior beliefs, in 33 series, the team that was leading 3-0 won the series. We also find no evidence that winning the fourth game and sweeping the series has an impact on the outcome of subsequent series. However, we find that leading team is significantly more likely to win Game Four. After adjusting for the impact of close game three, we find that the leading team won the fourth game 19 out of 21 times. We can reject the hypothesis that the probability of the leading team winning Game Four is 71% in favor of the alternate hypothesis of probability of winning greater that 71% with 95% confidence. Failure to reject the null hypothesis provides strong evidence against Gentleman's sweep. Our results suggests that NBA teams would rather win Game Four in spite of the direct financial incentive and the extremely low risk of prolonging the series. We conjecture that since the bulk of the NBA revenue comes from TV, merchandising and licensing, the marginal revenue from an additional game does not justify the perceived risks.