A Few Thoughts About 39

by A.J. Coltrane

This SI piece talks about Rick Peterson, then of the A’s, managing the pitching staff to try to avoid the 39th batter in a game. (That’s the cleanup hitter’s 5th plate appearance.)

…Oakland’s Opening Day rotation that year had four lefthanded starters, and Peterson learned that it was best to use his righthanded-specialist, Chad Bradford, as a preemptive strike against tough righty hitters even before his starter was exhausted. The reasoning was often to avoid matchups two or three innings later as much as it was about a particular at-bat in the present.

Peterson asked the team’s analytics department to research the correlation of winning percentage with the number of batters faced in a game. That research, he said, found a tipping point between 38 and 39 batters faced.

“Once they came back with that information,” said Peterson, who advocates for the use of biomechanics and advanced analytics in pitching through his company 3P Sports, “that answered my question. You’ve got to manage your bullpen [because] it’s critical that the 3-hole hitter doesn’t come up for the fifth time.”

Here’s why: Since 1991 home teams that have faced fewer than 39 opposing batters in a nine-inning game — four full times through the lineup, plus three additional hitters — win roughly three-quarters of the time (74 percent) while teams that have faced 39 or more hitters have won only 31 percent of games.

Moreover, in the last 22 seasons home teams that have faced 39 opposing hitters have won almost exactly 50 percent of their games — 50.082 percent, to be more precise — making 39 the inflection point of winning or losing.

The piece is accompanied by this table:

Winning Percentage As Correlated With Batters Faced 1991-2012

Now, is that really useful? Possibly to the pitching coach, though I’d think that there are a lot of other variables associated with it that make that observation fairly useless in practice.

Of course, the direct way to determine win probability is just to look at the scoreboard. Factoring in the runners on base, inning, and the out situation gives win probability, for an example check out the excellent post-game wrapups at Lookout Landing.

..and here’s a Win Probability Inquirer at the Hardball Times. That might a good link to save to the mobile device…

As a fan though, wouldn’t it be great to have a simple rule to know when the game is likely “over” without needing the assistance of an app? Maybe a high certainty to the results could be had with something like the following statement:  “If at any time during the game a team’s leadoff hitter has two more plate appearances than the opposition’s leadoff hitter, then the game has been decided at a 85% confidence level.” then “At three more plate appearances the certainty is 97%.”

For that matter it could be “If at any point after the 3rd inning one team has sent 10 more batters to the plate than their opponent then they’re 95% likely to win.” Maybe the number is 15. Maybe it’s 7… Maybe it’s the 5th inning and 8 batters…

I’m sure somebody knows this stuff, right?