José Berríos Is Terrible. Or Great. It Depends on How You’re Counting.

José Berríos has gotten shelled this year. Through five starts, he’s allowed 17 runs, 15 of them earned, good for a 4.71 ERA. Per our calculation of RA9-WAR, that means Berríos has been almost exactly replacement level, worth 0.1 wins above replacement so far this season. That follows last year’s debacle, when he was worth 0.2 wins below replacement by the same calculation. For a guy the Jays saw as their long-term ace a few years ago, it’s been a precipitous fall.

José Berríos has been lights out this year. He’s striking out 26.1% of his opponents and walking only 4.3%. That 21.7% gap between strikeout and walk rates is 15th among starters this year, just ahead of Gerrit Cole, who you’ve maybe heard of. It’s not just strikeouts and walks, either: Berríos has allowed only a single home run all year. He sports a 2.32 FIP. By our calculation of FIP-based WAR, he’s the eighth-best starter in baseball this season, just a hair behind Shohei Ohtani.

That gap between ERA and FIP is, to put it mildly, extreme. It’s the second-largest gap in baseball behind Nathan Eovaldi, who’s allowing a .413 BABIP so far this year – oof. What gives with Berríos? Let’s investigate and see which side feels more like the truth.

The first place I like to look when someone is putting up numbers this bifurcated is BABIP, but Berríos is running a reasonable .321 BABIP against. That’s high but not obscene, certainly not enough to drive a 2.5-point wedge between his ERA and FIP. The next place I like to look is quality of contact, because if you’re just giving up screaming missiles left and right, there are probably more doubles and triples against you than average. That often, but not always, leads to a higher BABIP, but it never hurts to double check.

Strangely, it doesn’t look like Berríos has had much of a problem with contact quality. He’s surrendering barrels (super hard-hit balls in the air) on 7.6% of opposing batted balls, much lower than last year and also lower than league average. He’s also keeping opposing hitters on the ground more often, with a career-high 49.4% groundball rate. That’s exceptional, and great for someone with Berríos’s skill set (read: limiting walks). Nothing to see here.

Okay, next question: Has the quality or mix of his pitches changed? He still throws his usual four-seam/sinker/curveball/changeup mix, so we’ll have to look at each one individually. First, the four-seamer:

Huh – his four-seamer looks better this year. He’s getting more vertical movement on it, which lets him throw it in the zone more often and garner more swinging strikes. Stuff+ thinks his fastball quality is mostly unchanged, for what it’s worth, and PitchingBot’s stuff metric thinks it’s slightly better than last year. Regardless of what you take from those, it’s clear that his fastball, which is either league average or slightly better when taking stuff and location into account, isn’t the main cause of his woes.

The sinker is similar:

The pitch is basically what it’s always been, a worm-burner. It’s running a ludicrous 60% groundball rate so far this year, which likely won’t continue, but it’s not missing many bats. That’s what Berríos’s sinker has always been; both stuff models concur on that point.

Curveball? Nope, not the culprit:

I really like what he’s doing with the curveball, throwing it out of the zone more frequently to induce chases. Stop me if you’ve heard this before: both stuff models think that the pitch is about as good as it’s been for years, and his best offering overall.

Finally, we come to the changeup. Surely, we’ll find the solution here, right? Well, maybe:

I’m frankly at a loss. He’s throwing his changeup slower, but velocity isn’t quite as much of an absolute good when it comes to changeups. He’s missing a ton of bats, but giving up more damage on contact, hence the middling pitch value per 100 pitches. And those darn pitching models are no help. Stuff+ thinks the pitch is much better this year thanks to the declining velocity, while PitchingBot rates it as roughly unchanged.

If you thought there was something weird about those numbers, you’re getting at Berríos’s biggest problem so far this year. Three of his four pitches are above average, and by more than his least-used pitch, the changeup, is below average. That’s not in some abstract stuff grade – that’s in results. As far as pitch values are concerned, a single is a single is a single; there’s no accounting for league average BABIP, no smoothing, none of that nonsense.

That’s because Berríos has been above average when it comes to the batting line he allows opposing hitters. Opponents are hitting .239/.278/.330 against him this year, good for a .271 wOBA. That’s basically Martín Maldonado’s career line, and a pitcher who could turn the opposition into a squad full of Martín Maldanados is what every team in baseball is hoping for. What the heck is going on?

Here’s another split for you:

Uhh… yeah, that seems bad. That’s how you end up letting more than half the runners who reach base against you score. Berríos’ 49% left-on-base rate is the lowest of his career by more than 20 percentage points (excluding his partial 2016 debut season). The major league average is 71.5%. When there are runners in scoring position, opposing hitters turn from Maldonado into Vladimir Guerrero Jr. but with more power. In two words: not great.

I looked at each individual pitch to see if Berríos is doing something different when runners are on base – perhaps letting the moment get to him and over-throwing or something of that nature. It’s there, if you’re willing to look really closely; his sinker is getting about an inch less sink and he’s throwing his changeup nearly two ticks harder, reverting to the old shape. Neither of those two are great, but they’re not enough to account for the comically large difference in results.

I continued looking for similar problems and continued to not really find anything. Berríos throws more sinkers with runners on base, which generally makes sense; he wants to induce grounders. He also throws more curves, looking for non-contact outs. This all seems like solid strategy to me. I watched a few videos of his mechanics and didn’t see much amiss, though that’s admittedly not my area of expertise.

I did find one thing, though. Berríos has a bit of a tendency that smart teams might be picking up on. He throws a ton of first-pitch curveballs with runners on base, more than half of his first pitches with runners on base this season. In total, he’s thrown more 0-0 curveballs with runners on than with the bases empty, despite facing more batters with no one on base. Opposing hitters seem to be sitting on it, and they’ve absolutely tattooed the pitch. They’ve taken eight absolutely monster hacks and been rewarded with three 100-plus-mph batted balls for their trouble. This is an out, but it’s not the kind of out you want to count on recording:

Knowing what your opponent might be throwing is a big advantage. But it’s not that big of an advantage; per Statcast, the 26 first-pitch curves that Berríos has thrown with runners on base have cost him 0.8 runs relative to average so far this year. That doesn’t explain his huge discrepancy between results depending on whether someone’s on base.

This is an unsatisfying answer, but I think that a lot of Berríos’s bad fortune has just been that – luck. I’m not a huge fan of his predictable pitch mix to start out at-bats when there are runners on base, and I hope he switches that up, but that’s small potatoes. He’s not throwing more meatballs, though, or hitting the edges less frequently. His swinging strike rate is higher with runners on, not lower. Opponents are just walloping the ball. Or, well, kind of… he’s actually allowed a lower xwOBA with runners in scoring position than with the bases empty. He’s just massively underperforming that number – a .471 BABIP is rarely earned, to say the least.

So, is José Berríos great or terrible? I think he’s somewhere in the middle. He probably won’t continue to post such an excellent FIP – if nothing else, his home run luck is bound to change. But he probably won’t keep being victimized with runners on base with nearly the same frequency that he has in the early going. To me, he looks like an above-average starter, just one having a nightmarish time whenever a runner reaches.