This pitcher has a 10.35 ERA  ... and a 0.85 ERA. Let's explain

Brayan Bello has a 10.35 ERA. He’s been the least effective starter in baseball. Earlier this season, it got him sent down to Triple-A, despite being relatively early in a six-year contract extension.

Brayan Bello has a 0.85 ERA. He’s been the most effective reliever in baseball. Yes, even more than Mason Miller, at least by ERA.

It's been a wildly down-then-up year for the Red Sox, one in which they lost their ace, dismissed their manager, went on the best stretch in team history, and, most recently, acquired a division foe's franchise player. There may be no better avatar for all of the incredible twists and turns of this group than Bello, who is in the midst of doing something that has never done before.

No one, ever, has been this bad (as a starter) and this great (as a reliever) in the same season. No one’s come close. (We’re looking at seasons where a pitcher had at least 30 innings in both roles, with at least five starts.)

Now: We know that ERA in small samples can be a little fluky, and that’s true here, too. No one really thinks that Bello is better than Miller. No one is really a true-talent 0.85 ERA pitcher for very long, and that’ll be true here soon enough, too. This list is not exactly a group of long-term luminaries.

But that ERA split is what’s happened, and so far, it’s the biggest disparity that’s ever happened. Even if the underlying numbers aren’t quite so extreme in either direction – Bello has a 3.34 expected ERA in relief, by Statcast’s reckoning, which takes into account the amount of contact as well as quality of contact, as compared to an 8.42 expected ERA as a starter – well, that’s still 5 whole runs worth of (expected) difference. If you want to chalk up the rest to good or bad luck, go ahead, but we’re still short a whole 5.00 ERA here.

So, then: How does one do that?

Perhaps better put: Is there something meaningfully different about starting vs. relieving? Or is it simply that August Bello is pitching better than April Bello did, and the innings where it comes are a coincidence?

As always, the best place to start is with the simple stuff, and nothing is more simple (yet powerful) than strikeouts and walks.

• As SP: 12% K // 11% BB // 2.5 HR/9
• As RP: 23% K // 4% BB // 0.17 HR/9

That's pretty stark: A strikeout rate twice as high coming on in relief; a walk rate cut nearly by two-thirds; a single homer allowed in 52 2/3 relief innings, as compared to 10 homers allowed in just 35 2/3 starter innings.

Changes of that magnitude aren’t “luck” or “fluky,” of course, because they can’t be. But nor is there any obvious smoking gun here, in terms of his pitch metrics.

Is he throwing harder? A little: 94.9 mph as a starter, 95.5 mph as a reliever.

Did he change his pitch mix? Somewhat? A little-used curveball was ditched; the cutter is seen less, the changeup is seen more.

Is he being more aggressive in the zone? Not really. An identical 47% zone rate in both roles.

Is he finding his true role? When asked recently if he liked relieving, Bello’s answer was, “It’s working.”

If none of that sounds earth-shattering, we would agree. Even something like Stuff+, which attempts to show how nasty, well, “the stuff” is (based on a model that works off of release point, velocity, movement, location) doesn’t see much here. The sinker is about identical. The sweeper grades out a little worse. The cutter and changeup are mildly better. It nets out to identical Stuff+.

Yet, obviously: Something has changed.

One thing that’s noticeable – if also confusing in its own right – is his arm angle, or if you prefer, how high his vertical release point is. After years in the 31° range, which is lower than the 37° Major League average, it soared north of 40° earlier this year, even into the 50s on some pitches. Throwing from a higher angle isn’t inherently good or bad, but it is different, and different is what we’re looking for right now.

It’s difficult to ignore this arm angle chart, which shows him going from quite high, back down nearer where it was, and then more recently back up …

… and how that mirrors the shape of his swing-and-miss chart, and that would be a tidy narrative ...

… except that the entire point here is how ineffective he was earlier in the year, so it’s hard to think it’s that – other than perhaps that this may point to some related but separate mechanical issues.

More importantly, perhaps, is something a subset of the Stuff+ model is saying. It’s not that his “stuff” is particularly nastier, and it’s not that he’s throwing it in the zone more, but it might be that the specific locations he is hitting are stronger. That’s what “Location+” is trying to express, where 100 is average, and Bello went from mildly below to excellent.

• As SP: 98
• As RP: 114

That’s going to encompass a whole lot of small things, like:

• As a SP, his four-seamer was in the non-competitive areas high above the plate 50% of the time; as a RP, that’s just 31%. Those higher pitches induced limited chases, and let batters write off the pitch entirely.

• As a SP, Bello's cutter was hitting the edges of the zone against lefties 37% of the time; as a RP, that’s 47%. The slugging percentage against has been cut by more than half from the massive 1.231 mark he’d allowed as a SP.

You can get that idea by looking at his vertical pitch locations against lefties – earlier in the year, he wasn’t pairing his two primary fastballs at all, making it easier to differentiate. Now, the four-seam and sinker are arriving much closer to one another.

Those are all small things, but small things can be meaningful ones. Yet is that simply a change that was going to be made following a demotion, as opposed to being specifically related to his role?

Finally, we get to the elephant in the room. Much has been made, certainly, of the first-inning struggles Bello had as a starter, where he allowed the sixth-highest first-inning OPS of any starter in the last 30 years. Maybe, then, it’s as simple as cutting out the first inning, allowing him time to get further prepared behind an opener.

But it’s not clearly that, either, because it wasn’t ever only that. In the fourth and fifth innings of games he started, he’d allowed a .761 slugging percentage, and while that isn’t the .837 (!) mark he allowed in the first inning, it’s not good, either. (That .761 is actually the second-highest of any starting pitcher in the fourth and fifth this year. The only pitcher with a worse mark, Simeon Woods Richardson, has been DFA’d by both Minnesota and Toronto this season.)

It’s not about having to not face batters more times through as a reliever, because while he did get hit hard in his third time through as a starter (1.149 OPS), it was nearly as bad the first time through (1.173 OPS).

Maybe, perhaps, there really is a there there, in that Bello benefits from not starting off with the top of the lineup, or having a few extra minutes to get going before having to come in. Yet if that were true, we might have seen it over the previous four seasons, and we mostly hadn’t – Bello had allowed a .790 first inning OPS as a starter previously, and a .717 OPS in all other innings. While that’s obviously a little worse, it’s also only mildly worse a spread than the average starter, who also fares worse in the first because the best hitters are stacked there.

The truth, as usual, is somewhere in the middle. In terms of true talent, Bello isn’t a 10-ERA starter, nor is he a sub-1 ERA reliever. Maybe the changes he’s made in relief would have come as a starter, too, and he’ll probably get a chance to be a starter again down the road. Maybe not this year, though.

In a rollercoaster year that very few teams have ever had, Bello’s having a year pretty much unlike anyone’s ever had, either.