Checking In On the Humble Changeup

The French writer Antoine de Saint-Exupery once said “perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.” One can go about interpreting such an assertion in any number of ways. When applied to baseball though, it reminds me of the changeup. In a game that, in recent years, increasingly prioritizes velocity and spin, the changeup’s characteristics feature the notable absence of those attributes. After all the changeup is, in essence, the absence of speed. That lesser velocity in turn corresponds to reduced spin. Interpreted through this lens, what makes a changeup is not what it is, so much as what it is not.

This post features seven visualizations, all of which aim to contextualize changeups thrown in 2020. They aim to first provide a picture of what changeups look like, highlight a few instances of outliers, and consider characteristics that correlate with changeup effectiveness.

To start, the following three histograms depict changeup usage, velocity, and spin, respectively. The data used come from Baseball Savant; only pitchers who faced 100 or more batters in 2020 and who threw a changeup at least 10% of the time are considered. These filters limit the group to 131 pitchers.

Those who rely on a changeup generally do so only intermittently. That said, using a pitch 15-19% of the time (the most likely case among those who throw it >10% of the time) isn’t insignificant. Still, a handful of pitchers were exceptions in 2020. Of pitchers with 100+ batters faced in 2020, NL Rookie of the Year Devin Williams used his changeup most often, deferring to it 52.6% of the time. Phillips Valdez and Zach Davies were the only other qualifying pitchers who used their changeups >40% of the time.

Changeup velocity resembles a loose normal distribution around the mid-80s. That said, changeup is obviously a relative term: Miguel Castro, Jonathan Hernandez, Jacob deGrom, and Sandy Alcantara all threw changeups over 90 mph on average in 2020.

An argument could be made that Devin Williams’ changeup should in fact be classified as a screwball given its considerable spin. With 2852 RPMs on average, it is far and away an outlier among changeups. On the other end of the spectrum, Sean Manaea‘s changeup averaged just 1104 RPMs in 2020. Most often though, changeup RPMs sit between 1500 and 2000; of the 131 pitchers in this group, 97 pitchers (74%) had a changeup within that range.

These first three histograms made a couple things clear. Changeups are most often not primary pitches but complementary offerings. They generally sit in the mid-80s and have spin rates below 2000 RPMs. Of course, there are always exceptions.

Next, a glance at changeup movement in 2020. Changeups are an interesting pitch given that they predominantly feature arm-side movement, often “tumbling” away from a pitcher’s throwing hand, in the direction of that hand’s pinkie rather than thumb. This horizontal movement makes changeups natural complements to sliders and cutters, which move in the opposite direction horizontally. Given the changeup’s arm-side trajectory, one can see the stark contrast between horizontal movement between right- and left-handed pitchers.

Left- and right-handed pitchers’ changeups move in different directions horizontally, but generally with a similar degree of movement in either direction. Of those pitchers in this group, the right-handed pitchers’ changeup featured 13.62 inches of sideways movement while those of the left-handed pitchers moved sideways 13.46 inches on average, in the opposite direction.

Below right- and left-handed changeups are evaluated together after taking the absolute value of each pitcher’s horizontal changeup movement in order to look at raw horizontal movement across all qualifying pitchers in the same quadrant.

Given this view, it does not appear that horizontal movement is correlated with vertical movement. In other words, average changeup depth does not tend to relate much to the side-to-side movement on that pitch.

The vertical and horizontal movement of changeups feature differing magnitudes. The mirror histogram below illustrates a couple points on this note. First, vertical movement (the drop) of changeups is greater than their horizontal movement. Second, the vertical movement across changeups features more variance, or is more widely dispersed, than horizontal movement.

Finally, a quick look at what changeup characteristics in 2020 most related to the success of those pitches. In order to evaluate success, I have made use of FanGraphs’ Pitcher Type Linear Weights (or Pitch Values), which aim to capture how effective a given pitch is. Changeup Pitch Values, or wCH/C, is juxtaposed with four changeup characteristics: average horizontal break, average vertical break, fastball and changeup velocity discrepancy, and changeup average spin rate.

This final chart comes in the form of a cross table, which offers the numerical and visual relationships between the multiple stats listed above, all in one place. The numbers offered in the upper-right hand portion offer correlations between the metrics. The scatterplots in the lower-left hand boxes illustrate how those points correspond to each other.

While no strong correlations appear above, some interesting ones do. First, changeup effectiveness, as measured by FanGraphs’ wCH/C, is most strongly correlated with spin rate, though that isn’t saying too much. Spin rate, meanwhile, correlates most to horizontal break. Interestingly, greater discrepancies between a pitcher’s fastball and changeup velocity as well as changeup depth (vertical movement) are not correlated with pitch value.

If horizontal movement even weakly relates to success, it is fair to assume pitchers will aim to maximize that type of movement on their changeups. The cross table above indicates that horizontal movement relates more to changeup success, on average, than the vertical drop typically associated with changeups (“drop off the table”). Devin Williams’ dominance in 2020 predominantly employing a pitch with elite horizontal movement alone will likely inspire imitators. To what degree players embrace that spin and whether or not they have any real success remains to be seen, though.