Bunting and the ghost runner: a causal inference approach
Abstract
In this paper, we investigate the effectiveness of the home team bunting in extra innings of Major League Baseball games when the game is tied in the bottom of the inning. Using methods rooted in causal inference, we show that teams choose not to bunt when the statistical evidence clearly suggests it is in their advantage to do so. The reluctance to bunt in this situation is likely tied to the general decline in bunting, which is considered less effective than swinging away in most situations. In the 2021-22 seasons, the home team's first batter bunted in only 21% of tied extra innings. When they bunted, the home team went on to win the game in 74% of those innings, compared to 57% when they swung away (i.e. did not bunt), for an odds ratio for winning comparing bunting to swinging away of 2.13 (95% C.I.: 1.13, 4.30). However, the bunters had more experience bunting and the nonbunters were stronger hitters, thus the odds ratio above is a confounded estimate of the effect of bunting. Using inverse probability weighting to adjust for confounding, we estimate an odds ratio of 1.86 (95% C.I. 1.07, 3.27). This means that a typical team would expect to win 2 more games each season if they bunted in such situations which is worth millions in player salary.
- Publication:
-
arXiv e-prints
- Pub Date:
- April 2024
- DOI:
- 10.48550/arXiv.2404.06587
- arXiv:
- arXiv:2404.06587
- Bibcode:
- 2024arXiv240406587C
- Keywords:
-
- Statistics - Applications