The Precedent for a Craig Kimbrel Changeup

As pitchers go, Craig Kimbrel’s been doing all right. Over the last three years, he leads baseball in adjusted ERA. “Well,” you say, “ERA can be misleading.” Absolutely so! Conveniently, over the last three years, he also leads baseball in adjusted FIP. “But,” you add, “FIP can be misleading in its own way.” Definitely. Observe, then, that, over the last three years, he also leads baseball in adjusted xFIP. For Craig Kimbrel, it’s been a clean sweep.

And he’s done it with two pitches: a really good fastball, and a really good curveball. Just using our pitch-type run values, the last three years, Kimbrel ranks second in fastball value per 100 pitches. He also ranks second in curveball value per 100 pitches. Usually you don’t want to read into this stat too much, because pitching sure is a complicated activity, but the right idea here is conveyed. Kimbrel has a good fastball, and he has a good breaking ball, and so he hasn’t needed anything else. He’s about as close to unhittable as a pitcher can humanly get.

With that in mind, this is delicious:

Meanwhile, Braves newcomer Jonny Gomes swung at some Kimbrel offerings and missed a few, including one that appeared to be a … changeup?
[…]
“Chris Johnson (another in the group that faced Kimbrel) told me he threw a couple of change-ups,” said Braves manager Fredi Gonzalez, who was watching pitchers on another field when Kimbrel threw.

Many are of the opinion that baseball has a strikeout problem. For a great number of fans, the ideal version of baseball is one in which there are more balls put into play. But it’s impossible to deny the appeal of watching a version of Craig Kimbrel with a game-ready changeup in his quiver. If sports are about watching humans push the limits of the possible, how good could Craig Kimbrel possibly be? Could he reach 60% strikeouts? This is the stuff of dreams within dreams.

Kimbrel, for whatever it’s worth, was quick to extinguish some of the enthusiasm:

“It’s nothing new,” Kimbrel said. “I joke around and play with [the changeup] every Spring Training.”
[…]
Kimbrel has not completely ruled out adding the changeup to his repertoire, which features an overpowering fastball and the knee-buckling curveball that he throws with the aggression of a slider. But for now, he simply views the changeup as a way to improve his mechanics and avoid putting too much early strain on his arm with an overabundance of curveballs.

In spring, Kimbrel throws changeups as he finds his mechanics. It’s safer than trying to find his mechanics while spinning breaking balls, and this allows him to find his proper arm speed, or so he feels. Many players will mess around with something in February and March before shelving the projects for the regular season, and it’s not like Kimbrel is going to have much opportunity to screw around in save situations. When the leverage goes up, a player’s likely to lean on what he trusts.

But you can see the door’s been left open. Kimbrel has messed around with a changeup for years. It stands to reason one will eventually work its way into a meaningful game. Could we see some changeups in 2015? Would there be any precedent for this?

We have pitch-type information stretching back to 2002. It isn’t of a uniform consistency — the further back you go, the worse the data presumably becomes. But it’s still definitely useful, flaws and all. So I checked on some historical offspeed rates.

I combined changeup rate and splitter rate, since the pitches act similarly, and can be confused for one another. As a sum, I called it offspeed rate. Last season, as you know, Kimbrel’s offspeed rate was 0.0%. He didn’t throw a single changeup. Now, between 2002 – 2013, there were 127 pitcher-seasons with an offspeed rate of 0.0%, where the pitcher threw at least 40 innings in Year 1, and at least 10 innings in Year 2. In Year 2, 51 of the pitchers posted an offspeed rate greater than 0.0%. That is, 51 of the pitchers threw at least one offspeed pitch, according to our data. Rafael Perez, for example, suddenly threw more than 20% offspeed pitches in 2010.

But then, Perez, in 2009, wasn’t very good. Thing about Kimbrel is he’s very good. So there’s not much driving him to change. I found 34 pitcher-seasons with an offspeed rate of 0.0%, where the pitcher threw at least 40 innings in Year 1, and at least 10 innings in Year 2, and Year 1 was fantastic. I defined “fantastic” as requiring an ERA-, FIP-, and xFIP- no greater than 80. In Year 2, in 14 of these cases did the pitcher throw at least one offspeed pitch. However, in 11 of those cases, the offspeed rate was below 1.0%. So there are only three such cases where a pitcher mixed in an offspeed pitch more than one out of every hundred times.

The history, according to our information:

• Alan Embree, 2002 – 2003, +2.0% offspeed
• Arthur Rhodes, 2002 – 2003, +5.0% offspeed
• Aroldis Chapman, 2013 – 2014, +6.7% offspeed

Embree was in his 30s, and in 2003 he was dealing with a little velocity loss. Rhodes was in his 30s, and in 2003 he was also dealing with a little velocity loss. Only Chapman could be said to have done this in his peak. Chapman didn’t need to change anything. Still, he did start throwing a few changeups, before putting it back on the shelf. Many of them went for balls, and Chapman didn’t seem to trust the location, but of the 19 swings hitters did take against Chapman’s changeup, 18 of them missed. Which is kind of exactly what you’d expect. Chapman had that weapon, and he didn’t even use it very much. He just didn’t need to.

So if you’re looking for some Kimbrel changeups in this coming regular season, Chapman establishes a history. Just last year, he busted some out, as a proven dominant closer on a team not going to the playoffs. But, probably, Kimbrel won’t throw any changeups, or he at least won’t throw more than a few. At this point in his career it isn’t necessary. It wouldn’t be a surprise to see him throw a changeup later on, if he sticks around in the game — that’s the transformation Francisco Rodriguez has gone through as a career closer. But as long as Kimbrel’s stuff is his stuff, he’s in no position of need.

As a fun side experiment, what might a Craig Kimbrel changeup look like? Using my pitch-comp system, I found some righties with similar fastballs, and then I averaged their own changeups. Based on those numbers, we might expect a Kimbrel changeup to go 89 or so miles per hour, with about eight inches of horizontal movement, and about five inches of vertical movement. This is, of course, a guess, based on something only a little better than randomness, but it’s probably in the ballpark. A good comp for that changeup would be Kelvin Herrera’s. Last year Herrera threw 18% changeups, with a 38% rate of whiffs per swing. For Herrera, the change was his main secondary pitch. Kimbrel, of course, would still lean on the curve.

That’s enough. That’s more than enough. Craig Kimbrel messes around with a changeup every spring training and it’s impossible not to daydream about. Based on the history, Kimbrel’s unlikely to feature a changeup in meaningful games until his performance starts to decline, but thanks to 2014 Aroldis Chapman, there’s a glimmer of hope we might see dominant Kimbrel mixing three pitches this coming summer. And if that were to happen, look out, world. With the game on the line in the ninth, your average pitcher won’t want to mess around. Chapman, though, was confident enough to play with his food. Craig Kimbrel is no less menacing a predator.