Blame the bullpen: Oakland A’s relievers get an F

After another loss on Tuesday night, the A’s are 14-27, the worst mark of any team in baseball. That record comes in spite of the fact that they’re fifth in the majors in runs scored (184), and have allowed "€œonly"€ 190 runs to their opponents; being outscored by six runs over 41 games generally does not lead to a disastrous win-loss record, but that’s where the A’s find themselves after the first quarter of the season. And as is often the case when a team’s record doesn’t match its run differential, the blame can be laid almost entirely at the feet of the relievers.

At the most basic level, you could just look at the A€’s 4.70 bullpen ERA — 28th in the majors — and conclude that they’ve struggled. But ERA is not a great tool to evaluate pitcher performances, and it’s especially poor at evaluating relief pitcher performance, because often their job is to come in and squash a rally; if they fail, the starting pitcher’s ERA is the one that goes up, so ERA won’t reflect bullpen performances in those situations.

And, of course, not all runs are created equal, especially late in the game. If you’re up 10-3 in the eighth inning and your bullpen gives up a few runs, it’s not a particularly big deal, as you’re extremely likely to win the game anyway. So when discussing a bullpen’s impact on a team’s record, we care not just about the number of runs they allow — whether ERA blames them for it or not — but also the distribution of when they allow those runs, and whether more of them happen to be coming in situations where giving up even a single run can have a dramatic outcome on the game.

Evan Scribner is a fantastic example of how different a reliever’s performance can be from his traditional numbers. If you look just at his overall line, it appears that he’s having an excellent season: 22 2/3 innings, 17 hits, 2 walks, 27 strikeouts, and a 2.38 ERA. Even newer-fangled pitching numbers like FIP (2.30) and xFIP (2.20) that are designed to better isolate a pitcher’s performance think very highly of Scribner. From a context-free standpoint, Scribner has been terrific. But once you look at the distribution of his performances, you find a somewhat different story.

In low-leverage situations, when the game is not really on the line, Scribner has been completely dominant, giving up just four hits to the 45 batters he’s faced in those situations, good for a .093 batting average. He’s faced 12 batters in high-leverage situations, and six of those have gotten hits, which is why he’s allowed 63% of his inherited runners to score. Those runs haven’t counted against his ERA, but they’ve had a direct negative impact on the A’s chances of winning. And that’s why, despite very good numbers overall, Scribner has racked up -0.5 Win Probability Added, which measures the change in a team’s win expectancy while a pitcher is on the mound. In total, Scribner has made the A’s more likely to lose after he exits games than they were before he entered.

In 41 games, the A’s bullpen has amassed a major league-worst -3.7 Win Probability Added, so this isn’t to lay all the blame at Scribner’s feet. His teammates in the bullpen have been just as bad in important situations and not nearly as good at other times, so he just got used as an example of how tricky it can be to evaluate the contributions of a reliever. If we just wanted to talk about a reliever pitching poorly, we could simply point at Fernando Abad, who has racked up -1.0 WPA on the strength of a 7.36 ERA. But while the A’s relievers have performed poorly overall, the most remarkable thing about their performance has been how dramatic a negative impact they’ve had on their team’s record; they are costing the A’s an average of about nine points of win expectancy each game.

By Win Probability Added — or lost, in this case — on a per-game basis, no team has ever been worse than the 1999 Kansas City Royals, who surrendered seven points of WPA per game, posting a -10.9 WPA for the season. The A’s are currently on pace for about -14 WPA over a full year, so if they keep blowing leads at this rate, they’ll have the least effective bullpen we’ve ever seen. Of course, they probably won’t keep blowing leads at this rate, and it’s often fairly easy to be on pace for the best or worst something in mid-May. But A’s fans wouldn’t be wrong to think that this bullpen has been more frustrating than any they’ve ever seen. No team has had to sit through a full season of relief performances this poor for an entire year.

Those ’99 Royals, though, come pretty close, and like this A’s bullpen, they mostly did it through being remarkably awful in clutch situations. Yeah, their closer, Jeff Montgomery, had a 6.84 ERA, which was atrocious even by 1999’s offensive standards, but their bullpen’s total ERA that season was just 14 percent worse than league average; the A’s current bullpen ERA is 29 percent worse than average, by comparison.

Using a linear regression, we can use a team’s ERA relative to the league average to come up with an expected WPA per game. Based on the A’s overall ERA, we’d expect them to be giving up about three points of WPA per game, about one-third of what they’re actually giving up. And yes, on the historical scale, the 1999 Royals also rate out as one of the biggest offenders. As based on their ERA, they should have surrendered only about one point of WPA per game, not seven points as they actually did.

But the biggest underachievers of all time, at least relative to their own ERA? That would be the 1979 Padres, who posted an ERA one point better than league average but still managed to rack up -8.8 WPA, or an average of five percentage points per game. While Rollie Fingers and John D’Acquisto did most of the damage, that bullpen featured a remarkable season from Steve Mura, who somehow managed to give up -1.4 WPA while posting a 3.08 ERA. You could almost say that Evan Scribner is currently pulling a Steve Mura.

The good news for the A’s (and the Indians, who are having the exact same problem) is that clutch performance is generally not very predictive, and this isn’t the kind of thing that should be expected to last the rest of the year. But given that the A’s are already 13 games under .500, they’re going to need some seriously great performances to dig out of the hole that their bullpen already dug for them. And if their bullpen keeps doing an impression of the 1999 Royals or the 1979 Padres for much longer, then their season is most definitely over.