Davey Johnson needs to stop bunting.
Sacrifice bunting usually doesn’t work. By that, I mean sacrificing an out for a base often makes a team less likely to score a run.
The following charts are from Fangraph’s Win Expectancy, a stat based on historical data–the likelihood a team will win a game given any particular situation. (A more detailed explanation can be found on Fangraph’s website here).
On Thursday afternoon, the Nationals unexpectedly found themselves in an extra innings game after Stephen Strasburg gave up a game-typing home run with two outs in the ninth. With the game tied, the Nationals got the lead off man on base in the 10th, 12th, and the 13th innings. Each time, Davey Johnson ordered the next batter to bunt.
Wilson Ramos walked to lead off the top of the 10th. Based on historical data, this gave the Nationals a 58.2% chance of winning the game. (The below chart from Fangraphs chart shows the win expectancy for the Cubs, the home team, which was 41.8% at that time).
Davey then ordered Tyler Moore to bunt pinch-runner Kurt Suzuki to second. The Fangraphs numbers show this reduced the Nationals win expectancy to 55.9%. Put simply, the stats show the Nationals were more likely to win the game with a runner on first and no outs than a runner on second and one out. In this situation, the next two batters failed to reach base, ending the inning.
Ryan Zimmerman led off the top of the 12th inning with a single. According to Fangraphs, this again put the Nationals win expectancy at 58.2%. Again, Davey ordered the next batter–this time David DeJesus–to bunt. Again, the Nationals win expectancy fell to 55.9%. Again, the next two batters failed to reach base, and again, the Nationals failed to score.
Denard Span led off the top of the 13th with a double. Davey again ordered a bunt. Bunting a runner to third base is far more defensible than bunting a runner to second. The historical numbers demonstrate that a runner on second and zero outs expectancy of 67.2%. With Span on third and one out, the win expectancy rose slightly to 67.4%.
In this case, the bunt worked. Chad Tracy “drove” the run home with a squibber up the first base line.
Win expectancy is a good tool, but each situation is fact specific. It always depends on the runner, the hitter, and the pitcher. In this case, it was probably wise to bunt because of Denard Span’s speed, which makes him more likely to score on a ground ball or sacrifice fly. This is exactly what happened. If a slow runner like Wilson Ramos were on third base that inning, he may not have scored. Similarly, it made sense to bunt with a slap-hitter like Lombardozzi at the plate. If Davey had ordered Harper or Werth to bunt there, I may have thrown my TV out the window.
In the 13th inning, Davey played it right. But if he had played the percentages in the 10th and the 12th, we may have never seen a 13th inning.