In 2008, the Oakland Athletics called up rookie reliever Brad Ziegler. Despite a fastball averaging 85 miles per hour, he found great success in the major leagues. After setting a record with 39 consecutive scoreless innings to start his career, Ziegler finished his 11 years with a 2.75 ERA. How he achieved such success despite a lack of velocity is clear when you look at his mechanics.
He would lean over, let his hips sink into his legs, and whip the ball out at a submarine arm angle. It is easy to see he had one of the lowest release points in the MLB. One result of such an absurd arm angle is that all his pitches’ axes had been shifted from what would be expected out of a normal arm slot. As shown by the spin axes in the video (via BallR, from the catcher’s perspective), his fastball had topspin and his slider had significant backspin. This made all of Ziegler’s pitches outliers relative to what major league hitters usually face. He may not have had the most electric stuff, but his ability to make batters uncomfortable is what made him one of the most effective relief pitchers over the past decade.
What makes Tyler Glasnow’s release point special is far harder to spot from the standard centerfield camera angle. At six foot eight with an athletic delivery, he releases the ball closer to batters than anyone in the major leagues. While arm angles affect the spin axes of the pitches coming out – vertical arm angles create more topspin/backspin, horizontal arm angles create more sidespin – a pitcher’s extension influences pitches’ “effective velocity”. In short, batters only have so much time to react to an incoming pitch; the closer the ball starts to the plate, the less time the batter has. So as hard as it would usually be to hit Tyler Glasnow’s stuff, each one of his pitches are effectively faster than the radar gun would indicate.
The three dimensional release point of a pitch goes a good amount of the way in telling you how a batter perceives the incoming pitches. The horizontal and vertical release points, all else being equal, also affect the angles at which the pitches approach the strike zone. With batters generally tuning their swings for the average incoming angle, we can generally assume abnormality is a plus. Extension intuitively seems more like a one way street. With a high extension meaning higher effective velocities, it is safe to assume pitchers are trying to release the ball as close to home plate as possible. With these assumptions in mind, here is an example metric estimating the quality of a release point.
Based on the contortion exhibited in the video above, release score seems to be doing something right. Max Scherzer, Aroldis Chapman, and Jose Berrios were examples of pitchers with excellent stuff and abnormal release points (all three were 95th percentile or higher). However, the other end of the spectrum had some interesting names too. Brad Hand, Shane Greene, Mike Soroka, and German Marquez were just some of the pitchers with negative release scores. This means along with having unremarkably average X-Z release points, they had poor extensions. They may all have elite stuff/command to compensate, or there may be an unintuitive advantage to the decreased effective velocities.
Release scores are obviously not perfect; rather a sketch of how a deception metric would be built. Then again, it solely uses the average ultimate release points of a pitcher. It does not address how the pitcher gets there. For example, Yusei Kikuchi and Ryan Borucki averaged nearly identical release points. Yet Kikuchi probably had a more deceptive motion due to his leg lift hitch and the hiding of the ball. The three dimensional release point is easily quantifiable, but deception is inherently rooted in the pitcher’s biomechanics.
Deception is a sort of special sauce on top of whatever command and stuff a pitcher has. It is hard to quantify and thus is often left as the X factor explaining residuals in catch all metrics. But as helpful as deception may be, it is rarely worth sacrificing other components of one’s game.
Although Kikuchi started the season with the abnormal mechanics, an adjustment was made toward the end of the year. On August 7th, the hitch was elongated and slowed. Two starts later it was gone altogether. He would hold these hitchless leg kick mechanics for the rest of the year. After an offseason working with Driveline and the Mariners, Kikuchi has simplified his arm mechanics as well in an attempt to improve consistency. His weird mechanics made him interesting, but was not worth sacrificing command, consistency, and velocity.
Kikuchi may not have been able to handle chasing absurdity, but there are numerous pitchers who make it work. Marcus Stroman’s timing games, Rich Hill occasionally dropping his arm angle, and Clayton Kershaw’s unique delivery are impressive because they can pull them off. Quantifying the difference deception makes is a difficult task far outside anything the public realm has truly tackled thus far. In these early stages, the best we can do is make metrics (like Release Score) and begin building understanding.
Better data means better metrics means better understanding. With large amounts of Kinatrax data, traits like Kikuchi’s hitch or hiding the ball could be quantified and studied. Until then, every bit of value should be squeezed out of what the three dimensional release points have to offer.
I realize your article was mainly focused on the obscurity of release points. It might be worth investigating which pitchers get unexpected movement than anticipated for given release points. The deception would be in how the batter expected the pitch to move vs. how it actually moved.
That is an a compelling argument, especially for guys like Josh Hader, but it has already been done by BaseballSavant. They compare a pitcher’s pitches against others of of the same pitch type, similar release point, and similar velocity. They call it “movement vs average”. Here’s the link to the 2019 four seam fastball leaderboard: https://baseballsavant.mlb.com/leaderboard/pitch-movement?year=2019&team=&min=q&pitch_type=FF&hand=&x=diff_x_hidden&z=diff_z_hidden
Thanks for the comment!