Every good bullpen is the same; every bad bullpen is bad in its own way
With Danny Farquhar’s loss yesterday to Baltimore, the M’s tied the historically hapless Athletics with 12 bullpen losses on the year, the most in the majors. M’s fans know well, of course, that the pen hasn’t been operating at peak effectiveness, and that it’s cost the M’s several wins, but nationally, the M’s bullpen struggles may fly under the radar because they’re within the normal range, whereas Oakland’s struggles really do seem pretty unique. Dave wrote a great article at Rob Neyer’s Just a Bit Outside site on the A’s, but I wanted to look at the myriad ways bullpens can become a problem by looking at the M’s and Marlins struggles, and how they differ from the A’s.
The job of the bullpen’s pretty clear: don’t let inherited runners score, and snuff out any potential comebacks before they occur. The best way to do this, it seems, is to pair high-strikeout guys with platoon specialists, and in so doing limit balls in play, hits and runs. This is all pretty obvious, but if we’re going to talk about how this can go wrong, we should probably start with the platonic ideal of a bullpen. In recent years, the closest thing to a platonic idea we’ve seen has been the Kansas City and Atlanta bullpens, who’ve combined high-K set-up and closers with some highly effective left-handed specialists, and given up very few runs as a result. You can make a go of it in other ways, and the M’s had a good run for a year with the likes of Roy Corcoran, who induced a lot of weak (ground ball) contact for a while, and then abruptly stopped doing so. To be consistently good, the Aroldis Chapman/Craig Kimbrel/Jordan Walden type is effective because you take a lot of lucky elements out of play. There’s no BABIP fluctuations to worry about with strikeouts, and high-velocity pitchers often run lower BABIPs and HR/FB ratios because they’re just much harder to barrel up. A great K:BB ratio and lower HRs produces a great FIP, and that’s generally correlated with good runs-on-the-board results.
But not always. As of today, the Marlins rank 3rd in baseball in bullpen FIP at 2.93, essentially tied with the 2nd place Astros (?). They’ve got the lowest HR rate of any team, and they pair that with the 7th best K:BB ratio. Because their home park, some of that low HR rate isn’t related to their performance, but even after park adjusting with FIP-, they rank 5th in baseball. They’ve been good.
Sort to ERA, though, and the Marlins rank 21st, and by WPA, they’re consorting with the M’s and A’s – they rank 28th. There are a number of somewhat complicated factors at work here and one big simple one. First, their balls in play have found holes, and that’s pushing up their ERA relative to the fielding-independent numbers. Partly as a result of that high BABIP, they’re terrible at stranding inherited runners – they rank 3rd worst in all of baseball with 43% of inherited runners scoring, and they’ve converted just 4 of the 12 save opportunities they’ve had. This points to the simple factor here: their closer just lost it. Steve Cishek has been a solid if non-traditional closer for a few years, and he’s generally been quite good. A high-GB guy, he’s limited home runs, but he’s not just a Corcoran clone: his K rate started around 25% in 2012 and hit 30% last year. From 2012 through 2014, Cishek ranked 5th in baseball in WPA, with 6.98, just ahead of Koji Uehara, Kenley Jansen and Aroldis Chapman. This year, he’s last in baseball, and it’s not even close. Other players have chipped in, of course. Bryan Morris really WAS more of a Corcoran clone, with a high GB rate covering over a sub-par K rate and bad walk rate, but it worked for him in Pittsburgh and again last year in Miami, until it suddenly stopped working in 2015.
Lessons: getting a good MLB closer out of nowhere happens somewhat often, but a bad couple of months from a closer can absolutely kill your WPA, dent your playoff chances, and piss off your fans. The Marlins have had some luck go against them, and they seem to have a good closer candidate now in AJ Ramos, so they won’t finish the year with this huge gap between ERA, FIP and WPA, but it illustrates that a great FIP doesn’t always translate into great results. This reminds me a bit of the Athletics of last year, whose bullpen was good overall, but suffered through the implosion of Jim Johnson’s closing career in April/May.
The Mariners are the opposite of the Marlins in many respects. The percentage of inherited runners that they’ve allowed to score is lower than league average, for example, and the percentage of save opportunities they’ve converted is dead on the league average. Their FIP isn’t pretty despite a low HR rate because they’ve walked far too many hitters – their 10.7% BB rate is 28th in baseball. But they’ve pitched around it, thanks in part to a good BABIP, and that’s left their ERA at a decent 3.60 – far below their FIP, and not something you’d associate with their low WPA, which ranks 26th.
Like the Marlins, the M’s have one big offender in the WPA department. Danny Farquhar’s WPA of -1.41 ranks second worst in baseball, distantly trailing Cishek’s. Farquhar’s strikeout rate has tumbled this year to under 19%. It was 28% last year, and nearly 35% just two seasons ago. To make matters worse, his HR/9 has gone from 0.32 in 2013 to 0.63 last year to 1.17 thus far in 2015. I’d worried a lot about a decline in velocity, but it’s movement that tells the story much better – Farquhar’s cutter really looked like a hard slider in previous years, but this year it’s behaving like a slow version of his fastball – the difference in horizontal movement from his four seam is less than 4″ now, down from 6.5″ or so last year, and the vertical drop is less than 2″ different. It’s a straight fastball now, and batters have responded by killing it. But he’s not a closer, so you’d think that the M’s could quarantine the damage a bit, especially given worsening movement on his cutter. That hasn’t happened yet, as Farquhar leads the team in bullpen IP.
The M’s have done all right in save opportunities, but the Fernando Rodney experience hasn’t been kind to the team’s WPA. He’s at -0.54, largely the result of back-to-back appearances early in the year. He blew a save in LA, taking a -0.8 hit in WPA just two days after a dramatic game in Oakland in which Fernando gave up *4* runs in the 9th to the A’s, sending the game to extras. But of course that just meant the M’s got to face Oakland’s historic bullpen, and thus Tyler Clippard served up a Nelson Cruz HR in the 10th. WPA isn’t predictive overall, and it really really isn’t predictive in a case like this, where two games exert such a strong pull on his overall number because we’re only 25% of the way through the season for a pitcher who’s only going to throw 60 IP or so. Rodney won’t keep putting up an awful WPA, especially if he keeps converting saves. The TEAM won’t continue to have an odd gap between ERA and WPA, because Farquhar’s eventually going to cede time to Carson Smith and the now-healthy Tom Wilhelmsen. Dominic Leone (2nd worst WPA) is in Tacoma. This is the kind of bad luck that’s bound to even out. But while that’s true, that’s not the only form of luck operating here. The M’s still haven’t really paid for their poor K:BB ratio, and if regression helps in one area, it can just as easily hurt in another. In some ways, the Marlins have it easier, because they isolated the big problem and can now wait for their solid FIP to result in better actual-runs-allowed.
Lesson: especially early in the year, little things can cascade and have a big impact on a team’s record. The bullpen has been extremely deep in recent years, and just the other day, the M’s were able to deal Yoervis Medina for big-league help. But a 15-day DL stint and a lack of confidence in Medina (ironically, Medina leads the M’s bullpen in WPA) meant that Lloyd McClendon felt he had to continue to use Farquhar. Poor starts from Taijuan Walker’s left the bullpen somewhat overtaxed, and pushed everyone’s workload up, and the offense has struggled against non-Oakland bullpens, meaning the M’s pen has had a lot on its shoulders. Despite decent runs-allowed stats and good work with inherited runners, they’ve given up critical runs in tie games. Not all of this is likely to last, but it’s a big reason the M’s playoff odds have been cut in half in a month, and a big reason why they trail the Angels by 2.5 games.
As I think you may be tired of reading, none of this is predictive. The Marlins and Mariners could finish with above-average bullpens, and could even fare well by WPA by the end of the year. More save opportunities for AJ Ramos, and more innings period for Carson Smith and the picture could change. But it’s kind of interesting how many ways a bullpen can struggle. The A’s have had the misfortune of hitting on all of them at once, as Dave mentioned – their ERA sucks, their FIP sucks, and the impact’s been even worse due to sequencing and leverage (when the pressure’s greatest, the A’s get worse). All of this is magnified early in the year, of course, but knowing that things may regress is very different from how it FEELS to watch the M’s, or how it must feel to Marlins fans when Cishek trots in. It’s often said that sabermetrics can’t get at emotions – they can’t illuminate what it’s like to watch a dominant pitcher, or what it feels like when a team wins. In this limited case, however, I think they do a decent job. You just need to know which numbers to look at. FIP isn’t great at emotion. WPA, on the other hand, is a wonderful proxy for angst.