The Brewers’ Amazing and Worrisome Bullpen

As we creep up on the beginning of May, the Milwaukee Brewers have the best record in baseball. At 19-7, they’ve thrust themselves squarely into playoff contention, even if the pre-season projections mostly saw them as a third wheel in a difficult division; their early success combined with the struggles of their direct competition have opened the door for the Brewers to make a real run at the postseason. As Jeff noted two weeks ago, it doesn’t even matter all that much that our projections still aren’t that bullish on their future performance, because the cushion they’ve created with a strong first month of the season gives them plenty of room to regress and still be in contention.

Which is a good thing, because there’s almost certainly some pretty harsh regression coming the Brewers direction; one of the core foundations of their strong start has been a remarkable performance from their bullpen.

BullpenWPA

The red line is the Brewers total Win Probability Added by their relievers during the first 26 games of the season. The Padres are a close second, but then there is an enormous drop-off after those two, with the third highest WPA from a bullpen unit (SFG) checking in at +2.46, while the fourth highest (BOS) is just +1.76, less than half of the Brewers’ +3.98 total bullpen WPA to date.

Let’s put that number in some context. A couple of years ago, the Orioles made an improbable playoff run based almost entirely on the remarkable performance of their relievers, who posted the highest single season team bullpen WPA in the 40 years of history that the metric covers; their +13.86 reliever WPA was actually more than +2 WPA better than the previous record holder, in fact. The 2012 Orioles bullpen was worth about an 8.5% win probability boost per game over the course of a full season.

The Brewers’ have amassed +3.98 bullpen WPA in 26 games, which works out to a 15.3% win probability boost each game; that’s 80% more impactful than the most impactful bullpen of all time. At their current pace, the Brewers would amass almost +25 WPA from their relievers; the 2012 Orioles were the first team to ever get past +12, and only 12 teams (out of 1,100) have managed to break +10.

Francisco Rodriguez is getting all the headlines, as he looks like vintage K-Rod racking up saves left and right, but this has really been a team effort. Beyond Rodriguez — who leads the majors in reliever WPA — the Brewers have two other pitchers in the top 10: lefties Zach Duke and Will Smith. Yes, that Zach Duke, the one who has spent his career as a pitch-to-contact swingman, has spent the first month of the season as a dominating setup guy. Somehow, at age-31, he’s decided to start blowing hitters away, and has posted a 34% strikeout rate over the first month of the season, and he’s not just being used against lefties either.

Between them, Dukes and Smith have thrown 33.2 innings and allowed a grand total of two runs, and neither of them had any effect on an outcome of the game. Dukes gave up both of the runs, allowing the Braves to increase their lead to 4-2 in Game 2 of the year, and then giving up a run to cut their lead over the Phillies to 6-4 a week later. They wouldn’t score again against the Braves, so the run proved meaningless, while his teammates shut down the Phillies for the next three innings and added four more insurance runs in an eventual blowout.

And it doesn’t just stop with the left-handers. Tyler Thornburg has been the primary right-handed setup guy for the team, and he’s allowed one run all season; the final (also inconsequential) run on that 5-2 Game 2 loss to the Braves. The Brewers four most used relievers this year have combined to allow a grand total of three runs, and all three came in games where the run proved to have no impact on the outcome. Four pitchers, 51.1 innings pitched, and no meaningful runs allowed. This is how you rack up +3.98 bullpen WPA in 26 games.

We can talk about Charlie Blackmon being on pace for a +12 WAR season as the most obvious sign that early season performances shouldn’t be extrapolated, but there is no performance anywhere in baseball that is further from expectation than the Brewers bullpen. Keep in mind that last year, the Brewers finished 28th in bullpen WPA, and their unstable relievers were one of the main reasons the team underachieved. In 2012, the Brewers were 27th in bullpen WPA. This has been the team’s achilles heel for a couple of years now, and now, it’s the area that is almost single handedly carrying them to the best record in baseball.

That isn’t meant to discount the rest of the talent on the team. Jonathan Lucroy may very well be a stealthy MVP candidate, given his combination of offense and impact behind the plate, and with Carlos Gomez, Ryan Braun, and Aramis Ramirez anchoring the offense, the guys are going to score some runs. Matt Garza was a solid off-season addition, and while the rotation lacks any kind of true ace, it also doesn’t have any batting practice pitching machines. This team is a lot better than last year’s stars-and-scrubs failure.

But this Brewers team is being carried by its bullpen, and more specifically, the bullpen’s total shutdown of opponents in important situations. Team wOBA allowed, by leverage situation:

Low leverage: .295 (10th)
Medium leverage: .271 (3rd)
High leverage: .198 (1st)

wOBA allowed in high leverage situations will naturally be lower, because that’s when the good relievers are put in the game, but this is not a sustainable performance. The Pirates posted the lowest team wOBA allowed in high leverage situations last year, and they came in at .243; the Braves were second lowest at .265. Even if Francisco Rodriguez really is the best closer in baseball, and Zach Duke has been reborn as a dominating one-inning bullpen weapon, and Smith and Thornburg are going to continue to pitch at a level near the game’s best left/right setup tandem, the Brewers bullpen would still be in for a huge performance regression, simply because of the distribution of when they’ve allowed opponents to get hits.

Once you factor in the reality that K-Rod isn’t going to keep pitching like Craig Kimbrel, and Zach Duke is probably still Zach Duke, and well, the May-September Brewers are going to have to find a new way to win baseball games. Maybe that means Jean Segura starts hitting again, or they find a real first baseman somewhere, or Khris Davis remembers that he’s supposed to get on base occasionally. There are areas for the Brewers to get better, so we shouldn’t just assume that everything except their bullpen is going to hold steady while that group regresses to the mean.

But the impact of the Brewers bullpen on their first 26 games can’t be understated, and Khris Davis would have to turn into Babe Ruth to offset the coming reliever regression. The Brewers current recipe for winning isn’t going to continue to work at the rate it has so far, and they’re going to have to find a new way to beat teams that doesn’t involve protecting every lead they ever get. The good news is that the wins they’ve already accumulated aren’t going to get revoked, and even if they play .500 ball the rest of the way, their April performance will keep them hanging around all year long.

But if you’re buying the Brewers as an elite team based on their April performance, just be aware that you’re betting on the sustainability of a team posting the greatest bullpen performance of all time, and getting there by approximately the All-Star break. I think I’ll take the under on that happening, and we’ll have to see how well the Brewers can play when their relievers remember than they’re human.



Print This Post



Dave is the Managing Editor of FanGraphs.


Sort by:   newest | oldest | most voted
Jonathan Judge
Member
Jonathan Judge
2 years 28 days ago

Dave, really well done.

My one comment for others is that while the bullpen’s WPA will almost certainly decline for the reasons Dave states, the Brewers bullpen is also a distinct #1 in xFIP- and #1 in SIERA. The results may be fluky good, but the process is not. So, while some of the close games will probably start going the other way, costing them a few games, the odds of this bullpen regressing anywhere near “the mean” in terms of peripherals seems remote.

In other words, I think the smart bet is that the Brewers will be able to keep a lot of these wins they have banked so far, going forward.

Ousy
Guest
Ousy
2 years 28 days ago

Which teams are at the bottom? Realize it’s not as extreme on that end but curious. Thanks!

SecondHandStore
Guest
SecondHandStore
2 years 28 days ago

Great read. I’m very interested to see what happens when Tom Gorzelanny is ready to return from the DL. Zach Duke has been a surprise. I thought he’d be Michael Gonzalez 2.0. I agree he’s not this good, but he might be good enough that they can’t rationalize dropping him. Wei-Chung Wang is clearly the weakest reliever in the pen, but they seem to really want to keep him. Everyone talks about completing a trade, but it’s the Pirates and I just don’t think that’s a reality.

One nit-picky thing: Brewers are currently getting the 7th best overall production out of 1B. Also, somewhat hilariously considering Mark Reynolds is getting majority of playing time, the Brewers have had the best defensive first basemen in the early season. That’s not going to hold for the rest of the season, but I think the need of a “real” first baseman is a non-issue.

oh Hal
Guest
oh Hal
2 years 28 days ago

Reynolds, Gennett, Ramirez, Davis and Braun are all supposedly between below average to awful defensively. It must be luck or magic.

SecondHandStore
Guest
SecondHandStore
2 years 28 days ago

I think with Braun people remember his horrendous year at third base and assume he’s awful. He’s actually worked himself into at worst an average defender.

I also think the reports of Gennett’s poor defense are outdated. Again he’s never going to be a great defender, but he’s been pretty good. DRS and UZR agree on both players.

Most of Reynolds’ defensive woes came at third base. He was always better as a first baseman which probably isn’t surprising. Watching him play this year has been a pleasant surprise. Again, he’s not a great defender but at the end of the year I bet he ends with positive marks in DRS and UZR.

Davis and Ramirez aren’t very good defenders, but Ramirez at least isn’t a liability. He’s actually been really good so far this year. That first year with the Brewers was by far his best defensive year ever so who knows with him. The jury is still out with Davis.

The Brewers currently lead all of baseball in FanGraphs’ DEF metric. I doubt that is going to last the whole season, but I don’t think it’s entirely a fluke either.

Dustin Rucinski
Guest
Dustin Rucinski
2 years 28 days ago

Gennett has looked really good so far and Braun is at worst average

Jsprech
Guest
Jsprech
2 years 28 days ago

I don’t think your comment regarding the Brewers’ first basemen is nit-picky in the slightest. Both Reynolds and Overbay have been batting reasonably well and playing competent defense.

JS
Guest
JS
2 years 28 days ago

The Brewers are just amazing. It’s nice to see them starting to get some recognition but most people don’t realize the pre-season projections for this team were just flat out wrong.

I’m sure next year they will conveniently be projected to win as many games as they end up winning this year though.

Steven
Guest
Steven
2 years 28 days ago

Zach Duke might not continue striking guys out at 12 per 9 inning rate, but watching him last night at the Cards game, he looks like he might be a fairly effective reliever this year.

He appears to be altering his release point (verified by Pitchf/x) where he’s dropping his arm occasionally. He also appears to have ditched his 4-seemer which sucked in the past and is working off his 2-seemer more often where he’s getting plenty of run to his arm side, and he’s throwing his slider and curveball (which is more of a slurve) much more often.

I think a combination of deception and being able to command different pitches that run so much in either direction will make him effective against righties and lefties this year.

Scott
Guest
Scott
2 years 28 days ago

Yes, I agree, the bullpen has been pitching out of their minds, but to bring up the last two years is irrelevant to 2014’s success. No Badenhop, no Mike Gonzalez, no Axford. I dont think they had a lefty in 2012 either. Yes, they will regress, but not fall off a cliff. These guys are a lot better than the players previously mentioned. The article also fails to point out April/May are the toughest 2 months of the schedule (road trips in June could be brutal too). Brewers have played ATL, PIT twice, STL twice, and BOS so far. Go Crew.

Ruki Motomiya
Member
Ruki Motomiya
2 years 28 days ago

K-Rod might not continue to be exactly Craig Kimbrel, but he should continue to be quite good: Career 2.65 ERA with only only 3 above 3.00 in his career (And one was his rookie season + only 3.03). He’s shown BABIP suppression, HR/FB suppression and LOB% suppression over 782.1 career innings, which is enough to suggest he may have some level of skills in those eras, and most importantly K-Rod has continued with his reduced walk rate of last year. If he can keep up a 7.3%/7.4% BB% with his career 29.6% K% or last year’s 28.0% K% he’ll do very good things. As for the ‘pen, it will certainly regress because even amazing teams in bullpen power wouldn’t do this good, but they have largely revamped the bullpen since last year so last year might not be particularly useful to look at.

Edgardo
Guest
Edgardo
2 years 28 days ago

I thought Jim Henderson was the primary setup guy and he certainly deserved mentioned with that improved K/9 and BB/9.

Urban Shocker
Guest
Urban Shocker
2 years 28 days ago

I really appreciate this post. I was looking yesterday for team reliever stats, but I was also looking for team defense stats to see if any conclusions could be drawn. It’s worth noting that the Brewers are showing up as a top 10 team defense as well, and Dave, I was wondering if you could weigh in on the sustainability of the Brewers overall performance in light of the defense and the bullpen.

Vision
Guest
Vision
2 years 27 days ago

Huge Cards fan, not worried about the Brewers just yet, but the wins are banked and they still count, so I respect the fact they have a punchers chance now.

If I’m a Brewers fan though, I wouldn’t seriously believe Duke is going to continue being this good, nor would I continue to think my 1B platoon is going to be decent. Both are areas you can upgrade in-season though.

Archibaldcrane
Guest
Archibaldcrane
2 years 27 days ago

The Brewers are 7th in baseball in 1B WAR – I have no idea why Mark Reynolds was available off the waiver wire for nothing this offseason, given how many teams could use his production at that spot. Hell, Lyle Overbay is even producing some value there.

FeslenR
Guest
FeslenR
2 years 27 days ago

Well, Reynolds has a poor batting average (which most fantasy leagues use as a cat) and not such a good BB/SO ratio. He’s doing well somehow, but it won’t last…and most who play fantasy baseball realize this.

Steven
Guest
Steven
2 years 27 days ago

Apparently real teams don’t play fantasy baseball… and most people who watch real teams play realize this.

Jason B
Guest
Jason B
2 years 27 days ago

And most teams know exactly who and what Mark Reynolds was and is…which was why he was so readily available this offseason.

Mr baseball
Guest
Mr baseball
2 years 27 days ago

It’s pretty obvious why most stat geeks that get hired on to MLB teams get let go after a few seasons. Collectively the stat geeks on the internet have done a great job of raising awareness of advanced metrics. They have done an excellent job of better quantifying legit performance and fluke performance. However the stat geeks struggle to add much more value than what you see above.

Nearly everyone is now aware that the Brewers aren’t this good. There will be regression. But to use the word worrisome In the headline is misguided. The brewers bullpen while getting unbelievably good results, and certain to face major regression, is still very good. Thornburg is elite. You can take your regression all the way back but I’ll note the obvious improvements and use more data, go ahead and use less data Dave. Will Smith is very good, Henderson is solid. KRod is solid to very good. A low changeup is about as good a pitch as you can throw when paired with a very well located fastball on the black.

The Brewers will regress, no duh! They will go from 120 win pace team to one closer to 92.

Stat geeks are terrible at predicting breakouts. They are good at validating them however. I’m sure your models will catch up someday.

Probably after the brewers sign Kendrys Morales.

SRB
Guest
SRB
2 years 27 days ago

Need to pad those page views by churning out new “content”!

I wish I could take back my bet from Vegas on K-Rod finishing the year with a 0.00 ERA and 80 Saves though. Boy do I feel dumb now!

Jason B
Guest
Jason B
2 years 27 days ago

“Stat geeks are terrible at predicting breakouts.”

So you don’t understand how statistical models or projection systems work. May want to check into a primer on the subject (not said snarkily, just may help you to better understand why models don’t typically predict huge breakouts).

“Need to pad those page views by churning out new “content”! ”

Yeah, it would be much better to just post new content a couple of times a month. Who wants to read interesting, free, relevant baseball content? Blah.

Mr baseball
Guest
Mr baseball
2 years 27 days ago

So you agree models are not good at predicting breakouts, what’s the problem? You agree with my point. The reason I made this statement is *precisely* because I do understand the models. They are close to useless predicting breakouts for individuals, nothing controversial here. Breakouts are not random events for the individual.

Weather forecasting has similar limitations. We can reasonably predict events but only within a parameter, a range of possibilities.

Our models do not possess enough information to make perfect predictions. That’s the point of my comments.

You seem to deny the simple fact models we currently use are poor predictors of individual breakouts. That’s why I prefer to use additional data the models omit, this data requires observation and investigation. Something you apparently need a primer in.

ecocd
Guest
ecocd
2 years 27 days ago

WPA can still be a bit misleading as a way to compare performance. The Brewers have played in an abnormal number of close games this year which I assume also means they’ve had the opportunity to rack up WPA.

An interesting analysis, which would take a gazillion hours, would be to measure “potential bullpen WPA” vs. “actual bullpen WPA” If every reliever went 3-up-3-down in every performance, what would their WPA have been? Then you can come up with a “WPA Converted” type of metric. It still wouldn’t be perfect, but it would be a way to scale it across all bullpens for a better picture of bullpen performance.

You could also try scaling the “WPA Converted” on a game-by-game basis to a common base and average those “Game-Equivalized WPA Converted” across the entire season.

Of course, after all of that, you might just end up with something 98% correlated to RA/9.

Utah Dave
Guest
Utah Dave
2 years 27 days ago

This reminds me of the 2013 Cardinals batting average with runners in scoring position. They were miles ahead of the next best team. I kept telling myself they had to regress. October came and they still had not regressed. I am not arguing the statistical validity of the theory, but maybe the Brewers’ bullpen is on a similar path. As a Pirates fan, I sure hope not. They have made us look silly so far this season. But I guess we’re not alone.

Bucky
Guest
Bucky
2 years 27 days ago

The article mentioned Tyler Thornburg, and I think his performance isn’t so outrageous given he is geared to be a starter so his fastball has gained 2-3 mph in relief as he doesn’t have to pace himself. Just last night Thornburg all but manhandled the Cards in his 2 innings using a 94 mph fastball and knee-bending breaking ball. Agreed THIS level of performance is not sustainable but the base skill set isn’t going away. It’s considered pretty likely that once Frankie takes a step back and if Henderson is back in a funk that Thornburg will get the closers role.

It’s also worth mentioning that Brandon Kintzler is a pretty fair relief pitcher and he has missed a lot of time in April. Not that one guy can compensate for several other guys regressing but it may soften the blow.

Patrick
Guest
Patrick
2 years 27 days ago

Dave, what would you make of the fact that the Brewers starters have thrown the most innings in baseball thus far? I feel like that has to have an effect on the bullpen’s amazing performance. Not only does it keep guys fresh, but it gives Roenicke the opportunity to use his best arms in high leverage situations (which will most positively impact WPA). If the starters continue to pitch deep into games, isn’t it reasonable to expect less regression from the bullpen?

wpDiscuz