Pitch Classifications and Baseball’s Poverty of Language

There is a fragment of a Virginia Woolf quote that has long stuck out to me. In her essay On Being Ill, Woolf states that “…to hinder the description of illness in literature, there is a poverty of the language.” In my interpretation, Woolf is describing the inadequate lexicon for defining illness and those afflictions which characterize it.

Far less consequentially, and in the less meaningful context of baseball, I find myself thinking there is a sort of poverty of language in commonly used pitch classifications. Despite the ever growing list of fastidiously recorded and meticulously interpreted datapoints (velocity, spin, extension, release points, movement, tilt, etc.) to better characterize each and every pitch thrown, baseball still has just a small handful of pitch classifications.

Of course when it comes to pitch classification in baseball, there is a lot to consider. First and foremost when making considerations or attempting to classify any given pitch, one should note the obvious fact that all pitchers are unique. The same pitch type can in fact appear very differently from one pitcher to the next, rendering pitch types potentially quite distorted when considering them as a whole.

There are some generalities for accepted pitch classes, though. Generally speaking, curveballs feature more drastic vertical break and less significant horizontal, or east-west, movement. Sliders are typically faster than curveballs, and break more horizontally, toward or away from where a batter stands. While both the slider and the curveball fit under the umbrella class of breaking balls, their profiles are considered fairly distinctive (when forgetting the term slurve, anyway).

Given the ever evolving task of bucketing distinct pitches by distinct pitchers into black and white classifications, I figured taking a look at how those two common breaking pitches differ between the pitchers that throw them both as well as on average across a broad group of pitchers.

To do so, data has been pulled and aggregated from Baseball Savant. Baseball Savant supplied pitch level data for all curveballs and sliders that were thrown by pitchers who employ each pitch fairly regularly (min. 250 pitches for each classification). Applying these filters for pitchers in 2021 gave way to a 42-pitcher (predominantly starters) dataset. With that pitch level data, pitch features were aggregated across pitchers and pitch classifications.

Below are the average slider metrics for 21 of the 42 pitchers evaluated. One can determine pretty quickly that, despite all being considered sliders, these pitches manifest themselves quite differently across individuals.

Despite each being considered a slider, Corbin Burnes‘ offering spins with nearly 3000 rpm’s on average while Josiah Gray‘s barely breaks 2000 rpm. Despite the reputation of sliders for sweeping, horizontal movement, the sliders of pitchers like Gray, Matt Harvey, JT Brubaker, and Kyle Freeland have minimal east-west bend, on average. Finally, the range of slider velocity among this small group of pitchers is >10mph.

For all the baseball world’s attempts to capture, quantify, and describe what players do, to use Woolf’s astute words, there seems to be a poverty of language in describing pitches while those same pitches differ so drastically – by whole feet in a game said to be about inches and increasingly is about smaller units of measurement than that.

For comparison’s sake, the following table features the summary statistics for those same 21 pitchers’ curveballs.

As one might have expected given the reputations of these pitches, curveballs appear to be slower on average across these 21 pitchers. Vertical movement also appears to be more pronounced in a downward direction. For the sliders listed in the previous table, vertical movement hovered around 0 with some positive figures and some negative; in the case of curveballs, only Mike Foltynewicz and Jhoulys Chacín threw that pitch on average with positive vertical movement.

Still, there is considerable variance in curveball characteristics from pitcher to pitcher on this list, particularly in vertical movement for a pitch with such a strong reputation for that feature.

Let’s step back now though. In order to better summarize the differences between these two pitch classifications across all 42 pitchers who threw 250+ of each in 2021, below is a table that aggregates a couple summary statistics for each pitch class. The difference is provided as sliders versus curveballs. Of significant note is that this table below lists release point and movement data in inches rather than feet.

On average across this particular set of pitchers, sliders were 6.5 mph faster than curveballs in 2021. However, vertical movement was more pronounced in curveballs, where the difference in sliders versus curveballs (subtracting a negative figure for vertical movement in curveballs, resulting in positive figures) was roughly 12 full inches on average. Curveballs also featured more spin on average: ~116 more rpm’s than sliders.

Clearly these classifications have some key differences that are supported by data along with their more qualitative descriptions and reputations. While I may complain about a lack of breadth in descriptors, on average there appears to be some clear differences between these two classes of pitches.

Things get more interesting though on a pitcher-specific level of granularity. Below is a table listing the differences in those metrics captured above for each pitcher considered in Baseball Savant’s dataset. You’ll find that, for some, the distinction between slider and curve is for more significant than for others.

Velocity and movement metrics have been highlighted on a gradient scale in orange in those cases where pitches differ for any particular pitcher more than is average across the whole group.

As one interesting example, Joe Musgrove stands out as someone whose sliders and curveballs do not appear to differ all so much in certain regards, by just a couple mph and a few dozen rpm’s. However, the shape of his curveball is distinct despite having similar spin/speed qualities as it bends vertically more than a foot more than the slider, on average.

In summation, pitches that are called the same thing in reality differ considerably across pitchers, to the point that one pitcher’s slider might more closely resemble another individual’s curveball. Even looking a pitchers individually, some gaps between slider and curveball characteristics are far more pronounced than others.

Despite the considerable variety between (and across) player pitch classes, there are clear reasons that things have been kept simple when it comes to definitions. For one, fewer classes enables less room for error/mislabelling in defining pitches. Keeping things simple allows for simple heuristics for each pitches profile. Still, as pitchers continue to experiment and test the boundaries of their arsenals, one begins to wonder if those classifications used today are sufficient in describing what these players can in fact do with the baseball.