You may have heard that the Reds are approaching their bullpen a bit differently than other teams this season. The Reds aren’t expected to be particularly good this season, and as such, they are a bit more free to experiment.
One recent game highlighted the new-school approach of manager Bryan Price. In the third inning of Monday’s game, Brandon Finnegan started the 3rd inning with a 5-run lead and proceeded to implode, loading the bases before walking the first run of the game in. With no outs recorded yet in the inning, Finnegan was primed to give up several more. Price pulled the trigger on a highly unusual move: He went to the bullpen in favor of Michael Lorenzen, one of his better bullpen arms.
This decision was lauded by quite a few writers and pundits, including those here at FanGraphs. Craig Edwards used it as the impetus for examining the overall usage in the Reds bullpen so far this year, and Ben Lindbergh and Jeff Sullivan called it out in their latest “Effectively Wild” episode. The emphasis, in both cases, was on the decision to bring Lorenzen into the game. Which was a great decision! It was weird! It was wonderful! Most importantly, it worked!
There’s another aspect of this Lorenzen appearance that shouldn’t go overlooked, though. After Lorenzen worked the 3rd inning with great success, he stayed on for the 4th inning, in which he maintained their 5-1 lead. He retired the side in order with 10 pitches in the 4th, having thrown 14 in the 3rd. The Reds tacked on another run in the top of the 5th, and Price stuck with Lorenzen again for the bottom of the inning, now with a 5-run cushion. Lorenzen, once again, set down the side in order, this time on just 8 pitches. With 32 total pitches on the day, Price elected to turn to a lesser pair of arms in Cody Reed and Wandy Peralta to finish out the game (although not before allowing Lorenzen to lead off the top of the 6th at the plate).
While the 3rd inning represented a quintessential high-leverage situation, the 4th contained much less leverage, and the 5th, still less. The numbers bear this out: In the third inning, Lorenzen faced three batters in situations commanding a Leverage Index of 2.68, 2.66, and 2.53. The total Leverage Index of these three batters was a whopping 7.87. By contrast, the total LI associated with Lorenzen’s work in the 4th and 5th innings was 2.40. The six outs Lorenzen got in those innings weren’t as important, cumulatively, as the least important hitter in the 3rd inning!
(Click the graph for an interactive version)
Price was rightfully lauded for bringing one of his best pitchers into one of the most critical moments of the game. That’s only half of the equation, though. Knowing when to take a key reliever out of the game, in the context of the season as a whole, is just as important as knowing when to put him in.
As Edwards rightly notes, Andrew Miller is only on pace for about 88 innings this season. Andrew Miller threw 74.1 IP last season, and it was the most he had ever thrown in his tenure as a full-time reliever. It’s not as though the Yankees or Indians were trying to limit his usage — it’s that a reliever, any reliever, has a limit to the number of innings (and more appropriately, the number of pitches) they can throw in a season without breaking down or losing their effectiveness.
The question, then, is how to maximize the value of these innings. Lorenzen threw 2 innings and 18 pitches that he, quite possibly, didn’t need to throw. He consumed 2.40 “units” of leverage in the process. The next day, he was (quite predictably) unavailable. Price, faced with a close/late game situation, had to throw Peralta in the 7th inning of a one-run game, where he retired the top of the Pirates’ lineup in order, but consumed 4.22 “units” of leverage — 75% more than Lorenzen did in those two innings the day before.
This isn’t to say that “perfect” bullpen usage is achievable. The nature of the game is to guess when the situation you’re faced with will be the most important in the remainder of the game, or the remainder of the series, or the remainder of the homestand. In some cases, a more important, later, closer, more tense situation will arise in the same game, and you’ll have used your most effective bullets. In other cases, you’ll have used a pitcher in a big spot one day, and he’ll be unavailable in an even bigger spot the next day. In still others, the team will go on a run of 4-5 close games in a row, and lesser parts are needed to fill the surplus of close/late innings.
(Click the graph for an interactive version)
But the concept of “perfect” bullpen usage must start with the recognition of constraints, and an approach that optimizes the total leverage that a pitcher can consume within those constraints. It’s not enough to pick the right person for the job when the job is hard; it’s also necessary to pick the right person for the job when the job is somewhat easier, so that the right person for the next hard job is available. Michael Lorenzen did the hard job, but he also did an easier one, and as a result, wasn’t available for the next hard job.
I'm an actuary who plays around with baseball data on the side. Brooklyn, NY. Twitter: @malkusm