The UFO Slider, and Its Supporting Cast, Makes the Giants’ Staff an Outlier

Tyler Rogers, easily the owner of the Statcast Era’s lowest vertical release point, throws a rising slider that’s rising even more this season. The positioning of his forearm at release means that a traditional curveball grip puts his thumb on top of the ball and the rest of his fingers underneath; as the ball rolls off of his hand, it creates backspin in addition to the sidespin more typical of a slider:

Tyler Rogers, 72mph Rising Slider. ? pic.twitter.com/1pRK3ybSaT

— Rob Friedman (@PitchingNinja) June 28, 2023

As you can imagine, the traditional fastball grip places his index and middle finger pointing towards five o’clock or 5:30 rather than a more typical two or three o’clock from a three-quarters arm slot. This results in arm-side sidespin, but also some extra drop, such that his fastball sinks more than his slider:

The Tyler Rogers four-seam fastball has more drop than the average MLB curveball. ? pic.twitter.com/dZIjz9DDI1

— Codify (@CodifyBaseball) April 7, 2023

Needless to say, I’ve found these two offerings to be among the most unique pitches in the majors this season by a couple of slightly different methods. As a result, the Giants were a confounding data point when I used my team-wide pitch-uniqueness model to estimate which pitching staffs roll out the widest array of “looks.”

The model in question utilizes Euclidean distance, which you probably used in middle school to find the distance between two coordinate pairs on a graph. But instead of an x-y pair, I used nine different inputs: horizontal and vertical movement, spin axis in three dimensions, spin rate, and release point in three dimensions.

Last week, when I was looking for the next sweeper, I used the model to compare every pitch (min. 100 tosses this season) to other offerings from the same team to see if any one team was testing out a particular kind of pitch. There might be some new cutter/slider variations, but for the most part, it seemed that the model was more adept at identifying teams that needed to diversify their pitch types than it was at identifying new pitches. For that exercise, I left out release point since it’s typically not used to classify different pitch types, but for the definitive ranking of pitch type diversity, it should be included.

Previously, I’ve kept lefties and righties separate since the model can confuse a lefty changeup and a righty slider for being similar (among other pairs of offerings). The inclusion of release point here should enable the model to differentiate between offerings, though it also means that lefty pitches will stick out more since there are fewer of them. That’s how it is in real life, though, so the following analysis includes both lefty and righty pitches.

As for how I actually utilized the model, I took the mean Euclidean distance of the three most similar offerings team-wide (a metric that I called TED for “three Euclidean distances”) for every pitch thrown at least 100 times thus far. Then, for each team, I took the mean of these means for a ranking of most diverse pitching staffs. Here are the top- and bottom-five teams (for context, the mean TED was 1.92):

I felt justified in my methods for the most part, since Tampa Bay — known varier of release points — ranked second-highest, and the Mets and Nationals — both of which I identified last week as having lots of pitchers with similar offerings — placed in the bottom five. But I wondered how much the aforementioned Rogers had to do with the Giants’ number-one ranking. Additionally, an outlier like Rogers had more sway for the Giants, since they only had 39 pitches that qualified, while the lowly Mets weren’t helped out as much by the uniqueness of Brooks Raley, their one lefty stalwart on staff, since they had 48 qualifiers.

So, I tried a new method. Among all pitches, the median TED was 1.82, so I ranked teams by the percentage of their qualifying pitches that were above the median (i.e., were above-average in terms of their uniqueness). I used median since the distribution of TEDs was right-skewed, and I used the percentage of qualifying pitches since each team had a different number of qualifiers. Here’s how that looked:

Predictably, the Giants dropped, but only by two spots. Taylor Rogers, Tyler’s brother and a lefty who throws from a three-quarters slot, has a sweeper that ranks 18th in TED (Tyler’s slider and sinker rank first and second, respectively) among the 1,291 qualifiers. The Giants have other lefties, but none throw a pitch with nearly that amount of horizontal break; the 14.7 inches Taylor’s sweeper averages dwarfs the 5.9 inches Alex Wood gets on his slider, the Giants’ next sweepiest. Wood and Sean Manaea’s lefty sliders also stand out, ranking 35th and 37th, respectively.

As you might have guessed by Taylor Rogers’ standout sweeper, the Giants achieved their place near the top of the TED rankings thanks to the largest standard deviation (and interquartile range, if you prefer to keep with median-centric measures) of horizontal movements among their qualifying pitches. They have a lot of offerings with little to no vertical movement, but all of them have horizontal separation, while their smaller clusters of pitches with similar horizontal break all have vertical separation:

Aside from Tyler Rogers’ UFO slider, the Giants owe a lot of their high ranking to their lefties. I expected lefties to stand out more, but the TED rankings weren’t just a list of teams with the most qualifying lefty pitches. In fact, of the five teams with the highest percentage of pitches above the median TED, only one (Cincinnati) ranked inside the top 10 in terms of the percentage of their qualifying pitches that were thrown by lefties, and they ranked ninth at 30.4%.

Only 20% of the Mariners’ qualifying pitches came from lefties, the lowest of the top five teams in median TED, and they ranked fifth in standard deviation of horizontal movement and first in one dimension of spin axis (sine of spin axis, which is unsurprisingly correlated with horizontal movement; the Giants placed second here). In other words, Seattle’s movement spread covers most of the same areas as San Francisco’s, even with fewer lefties:

The Padres, on the other hand, achieved their high TED ranking due to a wide array of release points. They were the only team to place in the top five in standard deviation of both horizontal and vertical release point:

Sidewinding lefty Tim Hill is certainly a big help here — his two fastballs are in the bottom right corner. Additionally, none of the Padres lefties throw from an overhand slot; if the mean release point lies somewhere in between the lefties and righties, overhand lefties will be closer to it, which would lower standard deviation. Such a measure of spread can be misleading in this way, and that’s why I opted to use so many other variables — their presence renders release point more of a demarcator for handedness.

You may be wondering where the Rays are in all of this; after all, they top the TED rankings. They have a lot of overhand lefties in addition to Pete Fairbanks, the most extreme overhander in the league, which penalizes their standard deviation. But a simple eye test indicates just how different their pitching looks are:

In addition to the outlier Fairbanks, the Rays have three right-handed sidearmers in Kevin Kelly, Trevor Kelley, and Ryan Thompson. But what fascinates me the most about their distribution of release points is that the three sidearmers throw from similar heights despite very different horizontal release points. The Kell(e)ys’ arms look similar at release, with slight bends at the elbow, but Trevor plants with his driving leg towards third base while Kevin steps directly toward home. While the Kell(e)ys stand at 6-foot-2, Thompson stands at 6-foot-5, but he achieves the same release height by keeping his arm completely straight (from left to right: Kelley, Kelly, Thompson):

If anything, the model might be underrating the Rays because of guys like this trio (even if Kelley is in the minors at the moment); they should be given extra points for having pitchers with different horizontal releases despite the same vertical release, something that could mess with a hitter’s height-based pitching delivery expectations. While that same numerical result can sometimes be achieved by just moving along the pitching rubber, at other times it indicates a real difference in delivery that only adds to deception.

One way we might capture some of this extra variability is by using arm angle, which controls for height, instead of or in addition to release point. Either way, improving the model seems like it would be a useful enterprise given that three of the top five teams by TED (the Rays, Mariners, and Padres) are in the top six in ERA and four are in the top 10, while three (the Mariners, Giants, and Rays) are in the top seven in FIP. The only one of the top five that doesn’t appear near the top of either of those categories is the Reds; their high TED may largely be due to their cadre of lefties, so another way of dealing with handedness is likely in store for the model’s future. Perhaps the Giants would be tops in both of those categories too if they could find someone to pair with Tyler Rogers’ vertical release.