Last week, I wrote about an interesting and amazing phenomenon: the fact that even though the A's are 40-26 and own the American League's best record, by few metrics does their record reflect the caliber of baseball they've played in the season's first 66 games.
To recap that story in a nutshell: Oakland has a historically good run differential, one that would lead Pythagorean measures and casual fans alike to believe the A's winning percentage should be hovering slightly over the formidable plateau of .700 — .709, to be exact. That would make for a 47-19 record, give or take a win.
Take a look at the schedule, and what you see fits into that narrative perfectly. Only three times all year have the A's lost by a margin of more than three runs. That's 4.5 percent of the regular season to date, essentially meaning that Oakland has been winning, tied, or within reasonable striking distance in 95.5 percent of its games. Unsurprisingly, the A's have played .635 baseball in those 63 contests.
But obviously, there's something going "wrong" somewhere, as wrong as anything can go when a team is 40-26 and in first place. But again, the A's could easily be 47-19, and the unit that has received the lion's share of the blame for that underperformance is the bullpen.
The link between the team's underperformance and bullpen issues seems tenuous right off the bat when Oakland's bullpen is examined using baseline stats. On the year, the Athletics' bullpen has posted an ERA of 2.53, good for the fourth-best in baseball. They've allowed an American League-best 154 walks, trailing only the Cardinals for the lowest free-pass total in baseball.
By luck or by talent, the A's have been helped by their relievers' BABIP, which currently sits at .265. They're pitching to contact, and doing it very well — that BABIP is third-best in baseball, thanks to a line-drive rate of 18.9 percent, seventh best across both leagues. And they can afford to pitch to contact, with an AL-best BB/9 of 2.79, trailing only San Francisco's 2.57.
An interesting aside that has almost nothing to do with how well Oakland's bullpen has pitched: the unit has received more run support than any other team in the American League, as the offense has gifted it a remarkable 4.43 RS/9; basically, on average, the A's score one run per every other inning thrown by a reliever.
Thanks largely to their strikeout numbers, the bullpen's SIERA is ninth in MLB. Essentially, the single most accurate predictor of ERA thinks this bullpen is above average, nothing spectacular. It does ding them simply for pitching in the Oakland Coliseum, the fairness of which is an entirely separate discussion. On the one hand, yes, they would give up more runs if they pitched at the Ballpark in Arlington 81 times a year. On the other, the A's play at the Coliseum and their roster is designed with that fact in mind. Regardless, SIERA rates the bullpen as "good," other metrics rate them as average, others as excellent.
How about the bottom line, though — whether or not the bullpen contributes positively to Oakland's win totals. In a sentence, it does, but to a degree more fitting of a very good team than a team that has played elite baseball. The A's bullpen is tied with the Yankees' for ninth in WPA MLB-wide. Their mark of 1.71 is a far cry from that of the Giants or Padres, with WPAs of 4.31 and 3.98, respectively, meaning that those teams' bullpens have earned them over two wins more than Oakland's bullpen has earned its club.
The A's are slightly above average in reliever -WPA; the bullpen has accrued the equivalent of 16.17 losses throughout the course of the season. But it's fifth in the league in +WPA, with a mark of 17.89. Different kinds of bullpens are good in different ways; Oakland's is volatile but still comes out in the black, while a bullpen like Cincinnati's, only marginally worse than that of the Athletics, doesn't do much harm or good either way but still comes out on top.
That's where the perception of the A's bullpen "having blown a bunch of games" comes in. Pitching in relief is about as thankless a job as there is on a baseball team, especially pitching with a lead. And given that the A's have held a lead in the late innings in a huge portion of their games so far, there have been ample opportunities to lose the confidence of the fan base. On more than on occasion, Jim Johnson, Luke Gregerson, and others have taken advantage.
Even in exceptional circumstances like those of Sean Doolittle, there's only so much praise someone can be given for not blowing a lead. The man has thrown 18.1 innings in May and June, and hasn't allowed a single earned run. He has allowed five hits since May 1. He's walked one batter. His recent performance has been beyond incredible, more good than Jim Johnson's 2014 has been bad, yet it has received, at most, half the attention that Johnson's struggles have.
Much of that is circumstantial, obviously, given the money Johnson is making and his track record of success. The point remains, though: the Athletics bullpen is very, very good. That said, the fact that it ranks ninth in WPA means that it's only "good," not elite. The A's have the best run differential in baseball; if they had the best bullpen (in terms of WPA), they'd hypothetically have won 2.6 additional games (that's simply replacing Oakland's bullpen WPA with top-ranked San Francisco's), in a rather loose interpretation of the stat itself.
Add 2.6 wins to 40 and you get 42.6, still well short of the Pythagorean projection of 46.85. Even if the A's had the best bullpen in the league — and they certainly don't need to given their offense and starting pitching — the difference between top-tier and top-third only accounts for about 40 percent of the margin between their Pythagorean record and their actual one. Give them the fifth-best bullpen WPA and it's even less; the point is simply that the A's would be significantly underachieving from a Pythagorean perspective even if their bullpen climbed into the top five for WPA.
What is accounting for the biggest portion of that margin between math and real life? Luck, mostly. Maybe we'll have some more content on the topic next week.