Pitching at Altitude, Part 5: Changeups & Splitters

Filed under:

Welcome to “Pitching at Altitude”, a six-part series where we’ll take a data-based approach to solving the near 30 year old dilemma that is pitching at mile high elevation.

Welcome to the fifth entry of my new data-based pitching series about the Colorado Rockies. After evaluating almost every pitcher on the roster during the extended “Crafting a Gameplan” series (which you can find here), this time we’re tackling a more widespread, fundamental topic: pitching at altitude. It’s no secret that this is a difficult thing to do, one the Rockies themselves have been grappling with for many years.

Even now, after 30 years, there doesn’t seem to be an obvious answer to pitching at high elevation. What we’re going to attempt to do in this series is use data and physics to truly understand the dynamics that make throwing a baseball at mile high elevation radically and uniquely different from any other place.

Here are the series’ six parts:

Part 5 is about off-speed pitches, those being changeups and splitters. If you haven’t checked out the other entries, I recommend you do so in chronological order, because we went over some important data concepts in them that will be used in most (if not all) chapters, and I wouldn’t want you to get lost. I will include small explanations, but these pieces are already very long to begin with. Without further ado, let’s continue, and I hope you enjoy it!

We’ve already looked at fastballs and breaking balls, and now we have the last significant group of pitch types to go over: off-speed pitches.

Disclaimer: for our study, we’ll limit the scope to “only” changeups and splitters. Technically speaking, knuckleballs and forkballs are also off-speed stuff, but they’re so rare and their sample sizes so small that they simply don’t make sense to include (and many of the principles that apply to changeups and splitters apply to forkballs too).

With that being said, let’s look at changes & splits. We’ll be going over their role in an arsenal, how to create movement on a changeup, what kinds of changeup types are there, and how cambios change with altitude — how their movement and interaction with a fastball changes, what profile suits elevation best, and so on.

Finally, we’ll have a separate for splitters, because while they do share similarities in just about every aspect with regular changeups, they also have many unique quirks that make them unique. Let’s begin.

Off-speed pitches are a very particular kind of offering, one that frequently relies heavily on deception and the interaction with the pitcher’s fastball to succeed. Now, all pitches benefit from deception, but fastballs and breaking balls don’t have the same direct need to deceive in most instances; fastballs can overpower batters with velocity, while breaking balls often feature huge movement.

Changeups exist in between this spectrum, as they’re not thrown as hard as fastballs (the average MLB changeup was around 86 MPH in 2022) but also don’t feature the same kind of movement as most breaking balls. When a changeup does have huge movement (see: Williams, Devin and his Airbender), it frequently becomes a dominant pitch. But most don’t move a ton compared to other pitches in their velocity range.

As such, the interaction between a pitcher’s fastball and changeup is hugely important in determining the quality of a pitcher’s changeup. This goes not just for velocity difference, not just for movement profile (your changeup’s shape should be similar to your fastball’s shape), but also for mechanical adjustments too.

One of the big mistakes young pitchers will make, for example, is slowing their arm down on or changing their posture in some way when throwing a changeup in order to take power and velocity off the baseball. High-level batters will pick that up rather quickly, and the ability to sell the changeup as a fastball with your body as you throw it is a crucial aspect of the deception that will make a change good.

A great offspeed pitch is a multi-faceted pitch and a true equalizer. It can be thrown for called strikes, it can fool batters and induce poor swings out of the zone, it can prevent hard contact, and it can provide a fantastic weapon for pitchers to combat opposite handed hitters.

But how exactly do we manipulate the baseball in order to throw a good changeup? How does a baseball spin? And is taking velocity off the ball really the most important thing?

You may have heard a lot of different theories about how to throw a changeup, or what a changeup needs to look like in order to be effective, but I’ll ask you to eliminate those from your memory for just a second here so we can begin from scratch.

Assuming we’re already doing the mechanical aspects well — same posture, same arm speed, etc, the most important aspect of throwing a good changeup is creating deviation from the fastball. There are many ways to do this, but at the core of the process, we’re looking to eliminate backspin from the pitch.

This will make sense to you almost immediately if you remember what we talked about with fastballs in parts 2-3. On heaters (four-seamers), we’re looking to use backspin as well as we can in order to create vertical movement thanks to how the ball will fight gravity, and here we’re trying to remove it so our changeup won’t fight against gravity.

We’ve all seen pitchers with changeups that seem to float instead of moving sharply down, right? More often than not, the problem with a changeup that has no depth is that the pitch has too much backspin, and as such stays up. Essentially, if on a four-seam we were trying to maximize Induced Vertical Break (IVB), on a changeup we’re trying to minimize IVB.

It’s important to remember that removing raw spin and removing backspin are not the same thing. For a quick example, we can look at one of the game’s premium changeup artists, former Red and now Mariners ace Luis Castillo. Castillo’s raw changeup spin is actually on the high end for a cambio, at around 1900 RPM (not much lower than his sinker), but he creates really good depth on it.

How does it he do it? Check out his release below:

Luis Castillo, 89mph Changeup Release/Spin (slow). pic.twitter.com/9udsjwC0sj

See how Castillo is side-spinning the baseball? That’s the key here. The way you remove IVB isn’t just by throwing the pitch slower and/or removing raw spin, but by introducing different elements of spin to that pitch and removing that backspin. Of course, inducing sidespin also means you’ll create horizontal run on the pitch, as we talked about in Pt. 4.

Removing backspin isn’t the only important aspect of throwing a good cambio. While there are many different ways to throw one (see: Cabrera, Edward and his 3-4 MPH difference between his fastball and changeup), most of the time you also want to take a decent amount of velocity off the baseball.

This is part of the deception aspect, of course, but the exact amount of velo you should take off depends for every pitcher. Some hurlers might get away with only a 5 MPH difference, some might have changeups that work better with a larger differential. As with every other pitch type, there is no universal answer.

A general rule, however, is that the more movement you can create, the less velo differential you need. Some of the game’s better changeup specialists, guys like Sandy Alcántara, Zack Greinke, Logan Webb, Edward Cabrera, etc, have small velo gaps, but their changeups are devastating because they have enough movement to make up for it.

Taking velo off the ball is one of the most complex and delicate aspects of pitch design. There are many ways to do it, and since what works for some guys may not work for others, and since knowing what works requires working with individuals, I won’t dive deep into this particular topic.

However, I would like to point out that changeup development is not as straightforward as breaking ball development. Consider the bulk of major league relievers —are there more fastball-changeup relievers or more fastball-breaking ball ones? The answer is obvious —fastball-breaking ball relievers are far more common, and it reinforces the point: changeups are not easy to develop.

A lot of times, certain athletes are naturally gifted at turning over a change. Pitchers who naturally lean towards pronation at release will often have an advantage, as their motor preference lends itself to creating the sidespin a changeup needs in order to move well.

That may be enough theory. Let’s now apply it and look at the different types of changeups you might see at the big league level, shall we?

The way we’re going to do this is by comparing different types of changeup profiles, and seeing how they stack up against each other. There are a lot of dynamics to take into account here: the results these changeups create, their transition to altitude, and more.

Let’s start out by examining the differences between a changeup profile built on velocity differential vs. a changeup profile built on movement differential.

First off, the changeups based on velo difference. You can also call this kind of changeup the “parachute change” if you wish, because it often looks like this:

Lucas Giolito, 81mph Changeup and 94mph Fastball, Overlay pic.twitter.com/y92PgZs6wP

This is the classic changeup profile, built on deception and that significant velocity gap compared to the pitcher’s heater, making the ball “float” as it approaches the plate and often inducing off-balance swings when it’s a good one.

But there are some changeups that look far different, such as this one down below from reigning NL Cy Young Sandy Alcántara:

Sandy Alcántara, 98mph Fastball and 93mph Changeup, Overlay. pic.twitter.com/2vvuQLavY4

That is a 93 MPH changeup, just 5 MPH off his fastball. This is not just about deception; hard changeups of this kind often overwhelm batters with a combination of velocity and movement, to the point where they can be thrown on their own, as an independent pitch that does not need perfect interaction with a heater to be effective.

Slow changeups and hard changeups tend to create different results; they don’t always belong to soft and hard throwers respectively. Hard-throwing White Sox ace Dylan Cease, for example, has an upper-70’s change that’s almost 20 MPH slower than his fastball, and Zack Greinke has long been famous for his changeup being almost as hard as his fastball, even as his velo has ended up in the 90 MPH range in recent years.

But what kind of results do these cambios produce?

(Run Value is the run impact of an event based on the runners on base, outs, ball and strike count. The higher, the more runs it will produce. wOBA stands for weighed on-base average; basically, OBP that takes the manner you reached base into account. xwOBA is expected wOBA, which takes walks, strikeouts and contact quality into account. CWS% is the percentage of all pitches thrown that result in a called strike or a swing-and-miss)

It’s not a perfect threshold, but considering that the average MLB fastball is about 93-94 MPH and you almost never ever see a fastball below 88 MPH, below 80 MPH is what I’d call a slow change and above 90 MPH is what I’d call a hard change. Here are some of my takeaways from that chart:

If you want to think about it in simple terms, slow changeups are going to create softer contact and more swings, but more of those batted balls will go in the air. Hard changeups won’t be chased quite as often and will be hit slightly harder, but they draw swings and misses and can be a go-to groundball weapon for a pitcher, in a way that very few other offerings can manage.

Before we get into how cambios perform at high altitude, I also wanted to write up a quick section about two of the main spin profiles you’ll see in big league changeups. First off, a classic, the sidespinner, as seen below with Rangers pitcher Dane Dunning:

Dane Dunning, Nasty 84mph Changeup. pic.twitter.com/kb2lpkgDRv

Throughout this series we’ve mentioned the concepts of spin axis/direction and spin efficiency, and we’ll be applying them here as well. Dunning’s changeup is thrown with a 2:30 axis, at roughly 1800 RPM, and you can really see the crazy horizontal movement he gets on it.

The key here is Dunning spins his changeup with 99% spin efficiency along with that steady 2:30 axis —in other words, as close to perfect sidespin as you can get. Sidearmers and pitchers with the ability to pronate well will usually favor these kinds of changeups.

On the other hand, you have a changeup like Joe Musgrove’s:

Joe Musgrove, Dirty 86mph Changeup. pic.twitter.com/DGy1vAynNg

The shape of that pitch looks drastically different to Dunning’s, doesn’t it? Musgrove’s seems to drop off a cliff, but he doesn’t have that same extreme armside run on it despite the fact that his raw spin on his changeup is about 2000 RPM, well above Dunning’s. What gives, then?

Well, a few things. First of all, Musgrove’s arm slot naturally creates a higher release point and, as such, a higher spin axis. Musgrove’s slowball tends to leave his hand at a 1:30 axis, a far cry from the 2:30 that Dunning achieves. The key here is this pitch’s spin efficiency and the seam effects it has.

Musgrove’s change has a spin efficiency of 67%, which means there is a lot more gyro spin on this pitch than on Dunning’s. This low spin efficiency is part of why Musgrove’s changeup drops like it does, along with seam-shifted wake (SSW), something we’ve mentioned a lot in this series.

While that cambio starts off spinning at a 1:30 axis, by the time it reaches the plate it has shifted to a 2:30 axis, the same as Dunning’s. This difference in axis paired with the added gyro spin (which makes gravity have a higher impact on ball flight) is what gives the pitch some of that extra late movement that seems to make it drop off the table.

The end result is interesting: despite the extremely different spin profiles, Dunning and Musgrove’s changeups drop about the same (33 inches) despite Musgrove’s higher raw spin rate and velo. Obviously, Dunning’s has a lot more horizontal movement, but armside run is not the end-all-be-all for changeup success.

Seam-effects changeups like Musgrove’s are ideal for supination-dominant pitchers. Because these athletes can often struggle to pronate, trying to force them into a sidespinning changeups that require extreme pronation at the end is often a bad idea. Instead, leaning into the cut and looking for a two-seam orientation on a changeup can be enough to create the slight armside movement and good depth a changeup needs.

Seeing all these changeup profiles having success really drives the point home: the interaction between changeup and fastball is crucial. Speaking of which, let’s see how changeups behave at Coors Field.

Throughout the years, a gazillion different opinions on how to pitch at altitude have popped up. Sinkers, sliders, fastballs at the knees in general, not walking hitters, and so on. What about changeups? We’ll we’re about to find out. As always, we have our trusted results charts to help us get a general view of performance, but there’s a lot more going on here.

(Run Value is the run impact of an event based on the runners on base, outs, ball and strike count. The higher, the more runs it will produce. wOBA stands for weighed on-base average; basically, OBP that takes the manner you reached base into account. xwOBA is expected wOBA, which takes walks, strikeouts and contact quality into account.)

Woof. If you remember from Pt. 4, curveballs and sliders had wOBAs below .300 at Coors, but that is not the case for changeups, which suffer the largest wOBA increase of any pitch type at altitude, along with some of most drastic decreases in chase and whiff rates. That Run Value is the third-worst, behind only four-seamers and sinkers, and the whiff rates pale in comparison with breaking balls.

But why is that? Sure, changeup movement changes at altitude, but deception plays a huge role in the success of this pitch type, so why can’t it carry over as well as a breaking ball can?

In my opinion, this is because a changeup at altitude suffers twice. It’s not just that changeup movement is diminished, it’s that fastball movement also suffers drastically.

Changeups usually don’t have the nastiness level of breaking balls —their success is built on deception and interaction with the fastball more than raw movement. And as we know, fastball movement changes a lot at altitude, with four-seamers losing about 3-4 inches of carry and everything losing armside run.

Coors Field reduces the movement difference between the typical heater and changeup, therefore making it easier for batters to make contact and not chase. This can be traced back to spin characteristics. Remember this chart from Pt. 4?

As I said all the way back in Pt. 1, the more a pitch’s movement is based on the Magnus Effect (high spin efficiency), the more it’s going to suffer at altitude, and that chart is a perfect way to picture it. Remember, the higher on the Y axis you go, the higher the spin efficiency.

Right at the top of the chart you have the vast majority of four-seamers (red), true spin sinkers (orange), slow curveballs (blue)... and most changeups (green), all pitches known to suffer at altitude. Towards the bottom you have sliders, cutters and hard curveballs, all pitches that tend to translate much better.

This is not to say that changeups can’t work at Coors Field! Deception is a real thing, and a very difficult factor to evaluate for pitch models. But there are some changeup rules, if you will, that I tend to consider for altitude:

Really, the thing to understand about changeups at altitude is they’re a tricky pitch to evaluate and predict success for, even more than at sea level, since the interaction between the change and the heater is drastically altered at high elevation. Deception, arm speed and command become even more important.

Some people consider splitters to be just another changeup type. I consider myself among those people, but if I’m writing a piece about changeups at altitude, splitters deserve their own section. Pure splitter and split-changes are becoming extremely popular across the game, and if you watched Japan’s pitching staff in the WBC this past spring, you can attest as to how devastating a good splitter can be.

Splitters have a reputation for being a “Coors-proof” offering, but is that actually true? How about we bring out our good ol’ results chart first?

Now, an important disclaimer here: there haven’t been a ton of splitters tossed at Coors Field since 2015 (not even 2000), so some of this could be noise and small sample size stuff. However, there are still some things we can learn from the chart:

So far, this is sounding like a great deal. In case you were still wondering, that Coors-proof reputation is mostly warranted —splitters perform very well at altitude. But why is that? Why do they perform so much better than regular changeups?

The answer to this question lies in the typical spin profile of a splitter when compared to a changeup. Watch this splitter release from Shohei Ohtani:

Shohei Ohtani, Splitter release/spin pic.twitter.com/I36IS78fw1

Doesn’t that look extremely different to the first release we saw in this piece, that of Luis Castillo’s side-spinning change? That Ohtani splitter averages around 1200 RPMs, and its spin efficiency is around the 60% range. Both numbers are way lower than Castillo’s changeup, and this applies to the average split-finger when compared to the average change.

The average splitter in the big leagues last year had 1404 RPMs — the average changeup had 1749 RPMs. The lower the spin rate, the more gravity affects the pitch’s movement, which is part of why splits are considered Coors-proof.

It’s not the only thing, however. In 2022, 334 of the 483 qualified changeups in MLB had a spin efficiency of 90% or higher, which is a rate of 69.1%. Splitters, however, don’t even sniff that mark. Only 18 of the 53 qualified splitters in 2022 MLB had a spin efficiency of 90% or higher, a rate of 34% — in other words, more or less half. This piece of information is a huge part of understanding why splitters seem to translate so well at altitude.

The average splitter relies far less on the Magnus Effect to create part of its movement than the average changeup. That is exactly why splitters usually have less run, why they’re thrown a bit harder, and why they’re such a good fit at altitude. Splits are functionally close in profile to that Joe Musgrove changeup we saw earlier, but with far less raw spin. They use gravity, seam-shifted wake and velocity to be effective, not perfect sidespin.

As we said on changeups — the closer a pitch gets to 100% spin efficiency, the more Coors will alter its properties. Splitters are a natural fit for altitude because their low spin, low efficiency, high velocity, north-south nature translates very well. In fact, you’ll often see splitters with low spin efficiency gain a few inches of depth at altitude, as has been the case with Kevin Gausman (70-75% spin eff.) in the past.

There’s a lot to say on changeups and splitters. If you ask people across the industry, my guess is most will tell you that cambios are the one pitch type that public data-based models (such as Stuff+) have the most trouble assessing. So much of a changeup success is built on the tough-to-measure skills of deception and tunneling, after all.

Here’s a fun piece of data I came across: from 2015-22, at Coors Field, changeups thrown at 90 MPH or above have a .358 wOBA against. Changeups thrown at 80 MPH or below have a... .358 wOBA against. The trends hold — harder changeups equal a bunch of groundballs, softer changeups equal softer contact.

One of the most important aspects of changeup development is figuring out the type of changeup a pitcher should be throwing, and this is where understanding a pitcher’s motor preference matters so much. Not everyone should throw a changeup the same way. Hell, some pitchers may not even need to throw one!

Natural pronators have a significant advantage when it comes to taking backspin off the ball and throwing excellent sidespinners, but supinators can often have issues with that kind of pitch —and that is why being open to the new ways to actively manipulate the ball, such as SSW and reverse gyro, is so important. Can’t fit a square peg into a round hole.

Splitters are often good fits for supinators too, but they shine particularly brightly for pitchers who may struggle to remove backspin from the ball. The aforementioned Kevin Gausman is a great example of that, as is Mariners starter Logan Gilbert. The long history of arm trouble that seems to follow split-finger users is something to keep in mind, however.

With this piece done, we’ve now gone over every major pitch type there is. We’ve seen how their movement is created, how to grade them, and how they translate to altitude across five parts. For Pt. 6, we’ll be wrapping up this series with a macro-oriented look at pitching at altitude —strategies, pitch combinations, arsenal coherence, and the kind of athletes I would favor. I hope to see you there!

★ ★ ★

Please keep in mind our Purple Row Community Guidelines when you’re commenting. Thanks!