Baseball’s Fastest Changeup

Do you know who throws the fastest fastball in all of baseball? Almost certainly! Maybe you don’t know it exactly — maybe you’re not sure whether Jordan Hicks counts while he’s recovering from Tommy John surgery, so you give it to Aroldis Chapman. Maybe you differentiate between starters and relievers and want to put Noah Syndergaard, Jacob deGrom, and Gerrit Cole in the conversation. For the most part, though, you know these names.

Let’s change it up slightly. Do you know who throws the hardest cutter? Again, there’s a decent chance you know that it’s Emmanuel Clase, though he, too, will miss the 2020 season. The second name on that list, though, is Michael Lorenzen, and even if you didn’t know he threw that particular pitch, you almost certainly knew that he throws hard and is good.

Why am I asking this? Because I’m setting up for a far stupider question: who throws the hardest changeup in baseball? Frustratingly, it’s also Syndergaard (minimum 100 changeups thrown). But he’s out for the season — the fastest active changeup in baseball, then! I can tell you with near-certainty that you didn’t guess it. Maybe you got the second-fastest — Jacob deGrom. Heck, maybe you got number four, Tommy Kahnle. I’m fairly confident, however, that you didn’t get number one: Orioles reliever Miguel Castro.

Now that I’ve told you that factoid, it’s time to anticipate your next question: so what? Why do we care who throws the fastest changeup in baseball? The fastest fastball is visceral; it’s as fast as anyone can make a baseball go, which is pretty clearly awesome. You can watch the fastest fastball and be impressed without caring about context. It’s just fast!

The fastest cutter isn’t quite the same, but a near-100 mph pitch with cut looks like witchcraft. Clase isn’t even that great, and he’s still a joy to watch. The fastest changeup, on the other hand, looks like this:

Clearly, a little more explanation is going to be necessary.

What makes a changeup good? It’s obviously not pure velo; if that was the only key, there would be far more hard changeups in the game. It’s not movement, at least not exclusively; Felix Hernandez dominated baseball for years with a straight-ish changeup that ate batters’ souls.

One thing I could do today is try to figure out this extremely difficult and thorny question. It’s a tricky one — how the pitch interacts with your fastball is of utmost importance, and not in a straightforward way. You can’t just measure the velocity gap between the two, or the movement gap, and call it a day.

That sounds like a lot of work, though. While it’s an answerable question, it’s hardly an easily answerable question. False leads and wrong turns abound in that field of research. Rather than try to solve that particular conundrum in one sitting, let’s just take a look at Castro.

First, let’s talk usage. Castro’s changeup has meaningful horizontal fade — 10.7 inches on average, in the top 3% of all pitchers in 2019. That makes it more or less a reverse slider; for a left-handed hitter, the pitch takes a down-and-away break that looks like a southpaw pitcher’s hard slider.

Given that his other secondary pitch is an actual slider, the real kind, and that sliders have severe platoon splits, Castro’s pitch usage is no surprise:

He’s a two-pitch pitcher, more or less. Those two pitches just happen to be different based on which batter he’s facing.

For the moment, let’s set aside right-handed hitters. Aside from a sneaky changeup here and there, he’s just sinker/slider, and that works well enough: since 2018, when Castro scrapped his four-seam fastball, he’s limited righties to a .292 wOBA and .310 xwOBA. Those aren’t elite reliever numbers, but they’re serviceable — the xwOBA is roughly league average for right/right matchups over that timeframe, and he was far better (.273 xwOBA) in 2019. We’re going to be talking changeups to lefties here.

Against lefties, Castro is a bit of a mess. More specifically, they’ve lit him up to the tune of a .337 wOBA and .350 xwOBA, miles worse than right-handed pitchers performed as a whole against left-handed hitters (roughly .325 on both over that time frame). So that’s it, right? Case closed. The changeup only gets used against lefties, Castro is bad against lefties, and so the hard changeup doesn’t work.

Yeah… it’s not quite that simple. It’s certainly true that Castro is bad against lefties, but that’s not because of his changeup. In fact, the pitch is the only thing keeping him from being absolutely unplayable against opposite-handed batters.

Consider, if you will, a lazy way of looking at pitch results. When Castro finished a plate appearance against a lefty by throwing a fastball, he got obliterated. We’re talking scorched earth, nothing-left-but-a-memory level destruction. Want the specifics? It’s a gory .413 wOBA allowed, which is still good news compared to a .430 xwOBA. It’s a rounding error away from lefties hitting like Mike Trout against his fastball. You know what you shouldn’t do as a pitcher? Turn your opponents into Trout.

But when he ends an at-bat with a changeup, the exact opposite happens. Lefties managed only a .226 wOBA against the pitch and a .252 xwOBA. In keeping with yesterday’s article, he’s turning opponents into career Jeff Mathis. Less Trout, more Mathis: that’s the way you get ahead as a pitcher.

That’s a rather simplistic analysis, unfortunately. Consider what we’d see if Castro only threw changeups when up in the count 0-2, and only threw them out of the zone. If the batter manages to take, the plate appearance continues. If he swings, it’s over, and Castro gets credit for the changeup strikeout. Clearly, we’ll need to do better than merely looking at pitches that end plate appearances.

Instead, I’m going to look at a modified version of our pitch weights. The idea is pretty simple: in that same 0-2 count example, the strikeout is “worth” the difference between batters’ average performance after 0-2 and the run value of a strikeout. A ball is worth the difference between 0-2 production and 1-2 production. In this way, we can at least control a little for count — if the pitch makes the batter’s life harder, the pitcher gets credit, and if it makes the batter’s life easier, the pitcher gets a demerit.

Broken down this way, the changeup is still excellent. Castro has thrown 399 changeups to lefties since 2018 and saved four runs relative to average. That doesn’t give him enough credit, even: that’s relative to the overall major league average, not the average production from lefty hitters against righty pitchers. It’s not just strikeouts, but also lovely pitches like this:

Okay, so that’s roughly 40% of Castro’s pitches to lefties. How about the fastballs, which are the vast majority of the rest? Yeesh. 571 fastballs, seven runs below average. That’s a massive hole in his game. Consider it this way, and please note that this math is terrible and don’t try this at home and I’m a professional and so on: a full season of pitches, for a starter, is roughly 3,000 pitches. At a league average clip, that’s something like 95 runs allowed. If you could somehow make those 3000 pitches out of Castro’s fastballs to lefties — and you can’t, stop thinking you can, quit it — that starter would instead allow 131 runs. It’s so much worse than average.

Now, take that previous comparison out of your head. It’s bad math, and pitches don’t really stack that way, and run values don’t stack that way at the extremes, and so on. Instead, focus on this:

Or this:

You don’t need my goofy math to understand: Castro’s fastball is bad. It’s particularly bad against left-handers, but it’s bad overall. Fastballs as a whole have negative pitch weights, but Castro’s is worse by a significant margin than that. That’s the long and short of it: Castro’s changeup is great, and despite that, it doesn’t do enough to offset his dang fastball.

Is there any solution to Castro’s Jekyll-and-Hyde nature? I mean, more changeups wouldn’t hurt. The pitch’s biggest drawback is its zone rate: it’s in the strike zone roughly a third of the time, which won’t work if he’s throwing more than half cambios. Some of that is a count issue, though: when the batter is ahead, he locates it in the strike zone 38.2% of the time, still low but maybe acceptable. And the fastball isn’t much better; it hits the zone only 44% of the time, without the benefit of bailout swings at would-be balls.

“Throw your best pitch more and for strikes” isn’t practical advice. Lots of pitchers would like to do that! But for Castro in particular, I think it’s worth a shot. Maybe it won’t work. Maybe he actually just needs to work on his fastball a little more, or transition back to a four-seam variant. But work with me here. Miguel Castro throws one of the fastest changeups in baseball. When batters swing at it, they come up empty a whopping 40% of the time. I don’t want him to go back to the lab and fix his fastball. I just want more of this, more often: