Did September performance determine Aaron Boone’s bullpen usage?

The true value of bullpens can be seen in October. Andrew Miller is now famous for his 2016 work, Brad Peacock and Charlie Morton dominated baseball in the Astros’ World Series run last year, and in general teams are asking less and less of their starters come playoff time.

The latest trend in this structural shift to single-inning work is the effect September performance has on playoff usage. The Brewers, Braves and Astros all left off some of their most-used relievers from their Division Series rosters over concerns of overuse and the resulting ineffectiveness. The strategy worked for Milwaukee and Houston, but did the Yankees try the same thing?

Essentially the theory behind this is that your focus in October should be on deploying your freshest - or rather, fresher - arms in the biggest moments. The question at hand is how do you determine who is fresher, who is gassed, and who is simply in a hot or cold streak? Relievers are incredibly volatile and so the predictive power of one month over another is spotty at best, but was bought into to some extent in 2018.

The best way I can tell to determine fatigue, aside from velocity changes, which can be a little funky for relievers, is with FIP, and more generally, K-BB%. Both metrics are determined by process: how often a reliever misses bats, what their command is like, and even factors in velocity to a certain extent, because we know greater velo brings more strikeouts. The “biggest spots” in a game are pretty easily found by leverage index, and you would expect that players who had the best September process - possibly indicating that they were fresher - would work the higher leverage moments.

Here’s how the Yankees did in the Wild Card and ALDS:

Since you’d expect low FIP to match with high Li, we’d like to see a cluster in the upper left quadrant. There isn’t one. Ok, well, how about a more raw tracker of pitcher energy, K-BB%?

This time, we’d expect a high K-BB% to match with high Li, so a cluster should appear in the upper right quadrant. Like above, though, there’s nothing to indicate that September performance, and thus relative freshness, played a role in Aaron Boone’s decision making. The only pitcher whose work best reflected his deployment by leverage was Dellin Betances, and that was only due to coming into the Wild Card game in such a huge spot.

A lot of this probably has to do with Boone’s deserved criticism of leaving pitchers in too long. He let two of the three ALDS games get out of hand so by the time the “fresher” pitchers were in, the leverage was so low they couldn’t maximize their timing. We’ll have to wait until next year to see if the Yankees commit to this bullpen trend, but with a sample size of five postseason games, for now it’s a no.