Tommy Kahnle’s Changeup Change

Earlier this week, Miguel Castro’s hard changeup caught my eye. It’s a weird, good pitch, and it’s thrown by a pitcher who might otherwise fade into the background. What’s more, he’s still bad against lefties despite a spectacular pitch for attacking them. About the only thing that made sense to me in the whole scenario was that Castro uses his changeup to attack lefties, the way right-handed pitchers are supposed to.

We’ll get to whether that’s true in a moment. First, let me introduce you to a righty pitcher who looks at this conventional wisdom — changeups to lefties, sliders to righties — and says eh, pass. Maybe not introduce you, actually, because he’s a notable pitcher on a marquee team, but at least alert you to his weirdness. Meet Tommy Kahnle, the man who throws his changeup when he shouldn’t.

As a rule, pitchers hate changeups to same-handed batters. Of all the pitches that righties threw to righties in 2019, only 7.1% were changeups or splitters (a splitter behaves almost exactly like a changeup, and pitch classification algorithms sometimes struggle to differentiate between the two, so for the remainder of this article I’ll be lumping both pitches together). On the other hand, they love them against lefties — 17.5% of right-to-left pitches were changeups. It’s pitching 101.

Kahnle surely took pitching 101; he just doesn’t seem to care. His changeup is his best offering, and he absolutely leans on it against lefties. 59.6% of his pitches to lefty batters in 2019 were changeups. It can’t even properly be called a secondary pitch; it’s just a primary pitch! Nothing to see there — a changeup-heavy pitcher throws a lot of changeups to opposite-handed batters. Where it gets interesting is when he faces righties. What does he do there, in the matchup his pitch wasn’t designed for? Why, he throws a changeup 44.2% of the time, of course.

He’s not alone in this weirdness — Héctor Neris and Tyler Clippard, just to name two, do similar things. But Kahnle interests me, because he wasn’t always this way. In 2017, he was spectacular. A 2.59 ERA, a 1.84 FIP, a Gerrit-Cole-facing-minor-leaguers 37.5% strikeout rate and a minuscule 6.6% walk rate — he was nothing short of dominant. That year, he threw a changeup to righties 14.7% of the time. Huh?

Let’s take a pause here and zoom way out. Are changeups thrown to right-handed batters by right-handed pitchers actually bad? Usage would tell you so, but as any analyst worth their salt knows, usage isn’t the end of the story. Usage would tell you that sacrifice bunts were the dominant strategy in baseball back in the day. Results matter.

Remember up above when I said that changeups were only 7.1% of righty/righty pitches in 2019? There are a lot of pitches in baseball — that 7.1% comes out to more than 20,000 changeups, which gives us a robust sample. Those pitches, perhaps unsurprisingly, didn’t really work; over the entire sample, they cost pitchers 77.7 runs relative to average, roughly 0.4 runs per 100 changeups. That understates the problem, too, because a righty/righty matchup isn’t “average.”

When we calculate our pitch values, we look at the league-wide production after each count. If we instead consider how right-handed pitchers fare against right-handed batters after each count, changeups look even worse, because right-handed pitchers should have an advantage on right-handed hitters. After adjusting for that, changeups were 113 runs below average, or more than half a run per 100 pitches. I don’t know a better way to say it than to say that throwing a same-handed changeup generally turns your good fortune into the batter’s good fortune.

Okay, so right/right changeups are a bad idea on the whole. They’re even worse than the numbers make it look, because pitchers with fringy changeups probably don’t throw the pitch to righties at all — this is a sample of the good changeups. To know whether there’s a true platoon split to the pitch, however, we’ll have to check in on how they do against lefties. If changeups are just bad pitches overall, the usage pattern still wouldn’t make sense.

Spoiler alert: of course changeups are good against lefties! In 42,804 pitches, changeups saved 62 runs relative to league average production, and 139 runs, or 0.3 runs per 100 pitches, after taking into account the platoon matchup. Changeups thrown by right-handed pitchers show a marked reverse platoon split — exactly what we’d expect to find.

Tommy Kahnle, of course, is a right-handed pitcher. Against lefties, his changeup is great. That makes sense — he’s a righty with a great changeup. To be precise, his changeup has saved 17 runs relative to righty/lefty matchup average over 816 tracked pitches, a whopping 2 runs per 100 pitches. Against righties, it saved 2.45 runs — total — over 673 pitches, or 0.36 runs per 100 pitches. It’s better against lefties, just like everyone else. Case closed, time to move on.

Only, it’s not time to move on. He still throws it all the time to right-handers, and it still adds value. That’s nothing to sneeze at. His alternatives, here, are to throw a fastball or a slider. In 2017, he leaned on fastballs to right-handers — 65% of the time — and split the balance of his pitches evenly between sliders and changeups. He still throws the fastball, naturally. There’s just one problem. Kahnle’s slider kind of stinks.

Consider this: Kahnle gets more whiffs with his changeup than with his slider — against righties. He gets more grounders with the changeup against righties. He locates it better. It should be no surprise, then, that the slider performs poorly in handedness-based pitch value. In 355 right/right sliders, Kahnle has cost himself 2.8 runs relative to average.

That’s the matchup where the pitch is supposed to be good! How does the slider fare against lefties? It’s hard to say, because after throwing four sliders to lefties in 2017, Kahnle hasn’t thrown another. That’s right: two years, zero breaking balls to left-handed batters. The pitch has a place against righties — hey, here, look at this, sometimes the ball can break away from you! — but at this point it’s mostly a show-me pitch. The Yankees are no dopes — it’s hardly a surprise that they’re getting one of their best pitchers to throw his best pitch more often.

To some extent, this is putting the cart before the horse. “Bag the slider, throw a changeup to a same-handed batter” is, generally speaking, terrible advice. You also have to get the pitch to do this:

To some extent, the way to have your changeup work against righties is, well, to have a good changeup. Kahnle’s changeup is deceptive, fast, and moves a ton. That’s a good start. There’s more to it than just that, though. Against lefties, the go-to changeup starts middle-in before breaking away, ending up on the outer half or just off of the plate. That’s how Kahnle throws his:

He’s in the strike zone fairly often, 43.9% of the time. That compares to a league-wide average of 37.9%. Even with two strikes, he keeps it in the zone 39.9% of the time, while the league rate dips to 32%. That works out well, because Kahnle has a preposterously high whiff rate; lefties have come up empty on 48.5% of their swings against his changeup, which is as silly as it sounds. Something about the velocity and the tunnel with his fastball makes batters look foolish. The overall whiff-per-swing rate for right/left changeups is a mere 29.3%.

Simply porting that location over to right-handed batters wouldn’t work quite as well. Think about it from the hitter’s perspective. The spot on the plate that Kahnle favors against lefties would start out looking middle or middle-away to right-handed batters, a good pitch to swing at. Then it would break into the bat, doing the hitter’s work in terms of getting out front and pulling the ball. No thanks.

Instead, Kahnle has a slightly different approach. He generally starts the pitch further outside, often off the plate entirely. From there, it dips back over the zone, ending nearly six inches further towards the lefty batter’s box, on average, than his approach to lefties. For the hitter, that’s a less comfortable swing. To hit it, you’ll have to commit to swinging at a pitch that looks for all the world like it will end up off the plate, trusting the break to carry the ball into your bat:

Judging by his 42% whiff-per-swing rate on right/right changeups, hitters aren’t great at doing that. Instead, they bail too early, or read fastball and swing over it, and just make a general mess of things. Here’s the bad read:

And the swing-over:

As an added bonus, batters get so used to making decisions on pitches away that it lets him absolutely steal their souls with a cheeky inside pitch occasionally:

Is this fastball-and-changeup strategy sustainable? I have no clue. It can be done — Neris does it, and so does Oliver Drake, though Drake’s extreme mound positioning and delivery are an entirely different can of worms. They both throw deeper, less horizontal off-speed pitches, however — more up-and-down than left-to-right. The best comparisons might be Luis Castillo or Chris Paddack, but they both mix in plus breaking balls. Kahnle is using a pitch designed to crush lefties, and he’s using it to get righties out with little help from anything else.

Maybe batters will adjust with a year of data. Maybe Kahnle will gain confidence in his slider and head back toward his 2017 usage patterns. Most likely, I think, is that neither of those will happen. Tommy Kahnle will keep throwing a purpose-built lefty killer to righties, and he’ll keep getting away with it. As an avowed changeup junkie, I’m excited to watch him try. José Altuve, on the other hand, is less enthused: