The Importance of Being Modest (About Projecting Bullpens)

This offseason, the Cardinals went out and signed Andrew Miller. It was a new direction in Cardinals reliever signings — a multi-year contract for an established relief ace. They’d tried almost everything the prior year, from signing Greg Holland to a blowout one year deal to dipping into the bargain bin for Luke Gregerson and Bud Norris. They’d tried trading for Dominic Leone, tried promoting aggressively from within with Jordan Hicks, tried using depth starters from Triple-A in Austin Gomber and Daniel Ponce de Leon.

None of it worked. The 2018 Cardinals bullpen was capital-a Atrocious. The team got a whopping 0.6 WAR from their bullpen, fifth-worst in baseball. They had the sixth-lowest strikeout rate, third-highest walk rate, third-worst xFIP- (which controls for park and home run rate). The team missed the playoffs by 2.5 games and lost the Central by seven, and their bullpen was a whopping 6.1 WAR worse than the Brewers’ relievers.

The offseason brought Miller, but it also brought consternation. The Cardinals bullpen was atrocious last year, and it was losing Bud Norris, who had been a steady presence. Dakota Hudson was going to be starting, and he’d been one of the three highest-leverage relievers last year. Maybe Carlos Martinez could return to the bullpen if his injury woes kept him out of the rotation, but Luke Weaver was gone too — someone needed to fill innings.

The Opening Day bullpen looked kind of meh. Jordan Hicks and Miller held down the late innings, and Johns Brebbia and Gant were reliable, but Mike Mayers, Dominic Leone, and Alex Reyes were mysteries of various types. Here’s what’s happened since then: Hicks blew out his elbow, Miller has a 5.5 FIP, Reyes was wild and hurt, Mayers got outrighted off of the 40-man roster, Leone has missed most of the season with injury and has a 6.27 ERA. Carlos Martinez picked up a little slack by returning, which somewhat offsets Hicks, but yeesh -- basically only Brebbia and Gant have delivered anything like what the team wants.

If I told you all of those things before the season, you’d think the Cardinals had done it again, used a trash bullpen when if only Mozeliak cared a little for the team instead of the budget, they could have had an unbeatable unit. About that — the Cardinals legitimately have one of the best bullpens in baseball. They lead the NL in reliever WAR, are second in baseball in both strikeout rate and K-BB%, and have the second best park-adjusted ERA.

The Brewers bullpen is a tire fire. They’re a full 2 WAR behind the Cardinals by fWAR, and they’re underperforming their FIP significantly — by RA/9 WAR, the Cards are third in baseball and the Brewers 18th, separated by five wins. Josh Hader is striking out I kid you not 47.2% of the batters he faces, mid-season addition Drew Pomeranz is at 40.5%, and it just DOES. NOT. MATTER. Heck, of the Brewers’ 1.7 total RA-9 WAR, 1.2 has been provided by Adrian Houser, and he’s a starter now.

What does this mean, aside from the fact that the narrative that the Brewers are fading is funny in light of the bullpen turnaround worth 12 wins from 2018 to 2019? It means we probably aren’t as good as we think we are at projecting bullpens, even if we don’t think we’re that good. “Cardinals Nation,” a vague descriptor I generally avoid using, thought this roster would be awful, and the big free agent signing busted. Why aren’t they bad, and why aren’t the Brewers better?

First of all, we need to talk about last year’s bullpen. It was, indeed, awful. The best reliever by almost any measure was Brebbia. Hicks was up there, as was Martinez in a short stint, but just generally ew. Here’s what that analysis leaves out, though -- the back of the bullpen was the actual problem, not the front. Look at the relievers who provided negative WAR last year:

There’s no sugarcoating it — these relievers (and also overall great guy Jedd Gyorko) sunk the pitching staff in 2018. This wasn’t a case of no talent, or even a case of a missing top few guys, as Martinez, Hicks, Brebbia, and Norris were just fine. The team simply gave a lot of innings to players who performed poorly. Some of that is randomness -- Weaver, Ponce de Leon, and Gomber managed to be bad in relief and acceptable as starters. Some of it is the team’s insistence on having a lefty reliever to do important things, even when the options of a hurt Cecil, past-his-prime Lyons, and homer-happy Shreve clearly didn’t merit 64 innings of work.

This year, that mostly hasn’t happened. The team’s worst reliever by fWAR is somehow Andrew Miller (-0.4), but I’m totally okay with calling that a FIP small-sample fluke and moving on. Gant, Brebbia, and especially Giovanny Gallegos have been tremendous. Carlos Martinez would be more exciting as a starter, but he’s been fine as a closer, and Ryan Helsley has been great fun in a cameo.

How much of this should we have predicted before the season? How much of this could we have foreseen, if we only looked hard enough? Well, Gallegos came out of nowhere, but projection systems loved him. ZiPS thought he was the second-best reliever on the Cardinals (behind Miller) even as he missed the Opening Day roster. Brebbia’s ERA’s the past three seasons have been 2.44, 3.20, and 2.89 respectively. Maybe he’s real? His career FIP is 3.31, not exactly fluke territory. John Gant? He’s probably not this good, but it’s not nuts that a converted starter would get better in relief.

This isn’t some situation where a few guys have pitched out of their minds despite poor underlying talent, either. For the balance of the year, FanGraphs projects the Cardinals bullpen to be the second-best in the NL, seventh-best in baseball. The Brewers? They project only 0.1 WAR worse than the Cardinals. The total separation between the best and worst bullpen projections is only a win over the last month, but the ordinal ranking still shows which team projection systems think is best, and they believe in the Cardinals.

Since I love making broad sweeping generalizations out of small issues, let me say this: for my money, ascribing too much certainty of value to bullpens is the number one mistake that fans make when considering teams’ fortunes. I picked the Brewers to win a Wild Card spot this year, and I did it largely on the back of Christian Yelich and a great bullpen. In fairness, I picked the Cardinals to win the Central, but I’m a bit of a homer.

In retrospect, it wasn’t an awful prediction, but my reasoning was poor. I was just too sure the bullpen would be good again, too confident in a bullpen that got 76.2 innings of 1.29 ERA baseball from Jeremy Jeffress and nearly 50 of randomly 3.33 ERA from Jacob Barnes, who is kind of bad. I didn’t pay attention to the fact that Hader probably wasn’t a true-talent 3 win reliever. I let myself believe that they’d manufacture dominant relief innings even as Brandon Woodruff graduated to the starting lineup and Corey Knebel recovered from elbow surgery.

Projecting bullpens is hard! It’s so hard. You’d do just as well rolling dice — unless you’re predicting that the Yankees bullpen will be good or that the Nationals will trade for seventy-three relievers and still be bad. The r-squared of bullpen rankings from one year to the next has been a puny 0.177 over the past few years, which means that a bullpen’s value in one year only explains about 18% of next year’s value. Heck, the Mets traded for Edwin Jackson and have an awful bullpen, while the Red Sox got rid of Craig Kimbrel and massively improved.

Maybe the Cardinals bullpen will collapse the rest of the year. Maybe it will continue to be great. Maybe next year fans will go in thinking it will be amazing only to get Brewers’ed. It’s all a great crapshoot, though, which is kind of my point in writing this article. Be happy that the bullpen is great this year. Also be modest, though — even if you specifically didn’t think it would be a disaster this year, thinking things about bullpens is a great way to be wrong.