The Brewers have been in first place for basically the entire season, and their odds of making the playoffs exceed 90% — at least by one estimate.
But, the baseball fan can never be fully satisfied, and one thing to complain about has been the bullpen. After a dynamite April (2.45 ERA), the bullpen’s ERA ballooned in May (4.22), recovered somewhat in June (3.67), and tanked again in July (4.48).
The most commonly-heard hypothesis for this variance has been reliever overuse. The front office ordered Ron Roenicke to carry Rule 5 pick Wei-Chung Wang, even though he wasn’t major-league ready, in an effort to bolster the Brewers’ ailing farm system. Roenicke’s refusal to use Wang shifted some of his anticipated innings onto other relievers. Couple this with Roenicke’s insistence upon having an “eighth-inning guy” and a “closer,” regardless of matchups, and the groundwork is certainly there for some potential overuse.
Moreover, one reliever after another has gone through rough stretches, often coinciding with perceived periods of heavy use: Francisco Rodriguez (5.73 ERA in May), Will Smith (14.54 ERA in July), and Rob Wooten (7.88 ERA in June). The Brewers front office has somewhat fueled the perception of a worn-out staff, making no secret of their efforts to secure more and supposedly better relievers.
But the problem with getting excited about reliever inconsistency is that relievers are, by nature, inconsistent. As we’ve discussed before, relievers tend to have fewer reliable pitches, so when one of them breaks down, they are exposed fairly quickly. Relievers also pitch very few innings these days, meaning that one or two bad outings can take several weeks to smooth out. For these reasons and others, relievers are volatile by nature — that’s why they are not starting games, after all — and a reliever who remains consistent month after month is the exception to the rule.
The actual “problem” with the Brewers bullpen hasn’t been lack of talent or rest: it’s primarily bad luck. Let me explain.
Although we tend to measure relievers by their runs allowed, the more sophisticated way to measure their performance is by something called RE24: the effect that the reliever has on the opposing team’s run expectancy. The problem with using ERA for relievers is not simply that it is fluky, but that it often conceals outings in which a bad reliever allows inherited runs to score, but escapes before allowing any runs he himself is charged with. (Here’s looking at you, Mike Gonzalez). RE24 compares compare the number of runs that have scored (or are still likely to score, based on outs and base position) by the end of a reliever’s outing versus the number of runs that were likely to score before he started. The tabulation of the positive and negative outings over the course of the year tells us whether the reliever is actually “relieving” anything, as far as results go.
In March and April, the Brewers’ bullpen ranked 3rd in league-wide RE24; as of yesterday, they had fallen back to 14th. Has the bullpen lost its touch? No.
RE24 is a result, not a skill. As it turns out, the majority of a bullpen’s RE24 is explained by two factors: the SIERA (skill-interactive earned runs allowed) of a reliever, and the BABIP (batting average on balls in play) that the reliever allows. (r2=.54, p<.00001). These are two very different and important inputs. SIERA is the best runs-allowed estimator we have: it models strikeouts, walks, home runs, ground balls, and other factors that cause runs to occur, and it does so with excellent accuracy. In other words, it tells us how skillfully a pitcher actually pitched. BABIP, on the other hand, is fluky and prone to luck aberrations. Sometimes the baseball gets hit to where somebody is, and sometimes it doesn’t. BABIPs above league-average tend to go down, and the reverse is also true.
In April, the Brewers bullpen had the best SIERA in baseball at 2.33. As of yesterday, the Brewers bullpen SIERA ranked . . . secondin all of baseball (3.03) — better than the Mariners, Padres, or Braves. That’s right: the Brewers bullpen ranks second only to the Yankees in how well they have pitched, and that includes the innings of target practice thrown by Wei-Chung Wang.
The bullpen’s real problem over the last few months has been BABIP. In April, the relievers had a below-average BABIP of .274 (league average is .293). That wasn’t going to last. Sure enough, in May, the BABIP of Brewers relievers surged to .329. In June, it was .325. In July, it was .329. In August, it has finally plunged back to .258, and not surprisingly, calls to improve the bullpen have temporarily subsided. Regardless, the bullpen’s BABIP is now averaging .307 across the season, which is still likely to regress favorably in September, although that is certainly not guaranteed.
The flip-side of this fact is this: while the Brewers (and their fans) have been on the lookout for bullpen upgrades, it is unlikely that those upgrades would make much difference. Brewers relievers are already pitching about as well as anyone could reasonably expect, and counting on a reliever reinforcement to reduce BABIP is not a wise assumption. There certainly is no harm in trading up, particularly for a right-handed strikeout pitcher, but unless that reliever can simultaneously cover the left and right-side gaps from the mound, there’s only so much more any bullpen addition can do.
Finally, on the fatigue issue. I have some preliminary statistical findings on this, but they’re not quite ready for prime time. In the meantime, let me suggest that that if a team’s bullpen demonstrates the best pitching skills in baseball in April and has declined to merely the second-best in baseball by August, chances are this bullpen is not being overused or worn out. The bullpen in any event appears to on the upswing, and as the Brewers head into the home stretch of their pennant race, the Brewers’ relief staff appears poised to do their part.
Follow Jonathan on Twitter @bachlaw.
All statistics from Baseball Reference and Fangraphs.