An Arm Angle Update That Ends With a Mystery

Early final yr, I wrote two articles exploring the handful of pitchers who determined that, relying on the handedness of the batter they have been going through, they need to change not simply their pitch combine however one thing extra basic about themselves as pitchers. Some drastically lowered their arm angle in opposition to same-handed batters, whereas some scooched from one aspect of the rubber to the opposite. I principally wrote about these pitchers as a result of they have been enjoyable to observe, however I additionally dived into the reasoning behind their selections. It wasn’t onerous to grasp what they have been pondering: All issues being equal, throwing from a decrease arm angle works higher in opposition to same-handed batters, whereas the next arm angle works higher in opposition to opposite-handed hitters. I even had numbers to again it up. I ran correlation coefficients between the pitcher’s wOBA allowed and their launch level, and I used common velocity as a type of management variable.

The correlation coefficients on this desk are fairly small, however they point out that when the pitcher has the platoon benefit, vertical launch level issues an entire lot. In truth, in that pattern, it has the identical correlation to success as velocity, which is unquestionably a shock. When the batter has the platoon benefit, vertical launch level doesn’t have any bearing on their success, however horizontal launch level does. That’s why some pitchers scooch all the best way over to the alternative aspect of the rubber.

Now that you simply’re all caught up, it’s time to handle the large flaw in these numbers. The issue with my information was that I wasn’t truly utilizing the pitcher’s arm angle. I used to be utilizing their launch level – actually the spot within the air above the mound the place the ball leaves their hand – as a stand-in.

Right here’s what I wrote on the time:

Up to now, I’ve been utilizing vertical launch level as a proxy for arm angle, nevertheless it’s not that straightforward. Variations in top, wingspan, and mechanics imply that the equivalence between the 2 is imperfect. You may reverse engineer arm angle from PITCHf/x or Statcast information, however that may solely provide you with an estimate. I think about Statcast is able to monitoring arm angle. If that information ever turns into accessible to the general public, I feel it’d considerably change our understanding of the platoon benefit.

As chances are you’ll know, Statcast not too long ago made arm angle information accessible to the general public. Are you able to have your understanding of the platoon benefit considerably modified? Effectively that’s too dangerous, as a result of the outcomes aren’t notably surprising. Nevertheless, we are able to nonetheless glean a number of fascinating information. To start with, let’s replace our desk by including in arm angle. As a result of the information solely goes again to 2020, the pattern dimension is a bit smaller, which in flip makes the correlation coefficients smaller.

Regardless of the smaller pattern dimension, it’s reassuring to see that the traits are holding. Let’s begin with same-sided hitters. The next launch level, and one which’s nearer to the middle of the rubber, works higher in opposition to them. The brand new column tells us that in opposition to same-sided batters, when the pitcher has the platoon benefit, the precise arm angle is way more necessary than launch level, which is what we might have been anticipated. Slicing out these confounding variables makes it clear that we don’t want to make use of launch level as a proxy anymore; arm angle is what issues.

Now let’s flip to opposite-handed batters. When the batter has the platoon benefit, I used to be actually shocked to search out that horizontal launch level had a barely stronger correlation to wOBA than arm angle did. As soon as once more, these are extraordinarily small numbers, so we shouldn’t learn an excessive amount of into any of this, however if you happen to wished to, you might take this as proof that the pitchers who scooch all the best way over on the rubber in opposition to opposite-handed batters actually do have the precise concept. I attempted working a pair regression analyses to see whether or not combining horizontal launch level and arm angle resulted in a good stronger correlation, however that wasn’t the case. I think that horizontal launch level simply seems to be a very good proxy for arm angle, and that it’d even be factoring within the pitcher’s extension. I’d by no means ever discourage a pitcher from scooching over on the mound all recreation, however I’m unhappy to report that I not consider it might be that helpful.

I’d additionally prefer to carry up one curiosity. After I ran the numbers this time round I checked each wOBA and xwOBA, simply out of curiosity. In opposition to opposite-handed batters, the numbers weren’t notably notable. Nevertheless, one thing fascinating occurred in opposition to same-handed batters.

For some cause, each single metric had the next correlation to the anticipated wOBA than the precise wOBA. It moved arm angle previous velocity! I don’t have a very good rationalization for this. Pitchers with decrease arm angles carry out higher in opposition to same-handed batters, however their arm angle has a a lot stronger correlation to the anticipated wOBA of the batter than to the precise wOBA. Why might that be? Once we discuss in regards to the distinction between anticipated and precise stats, we’re simply speaking in regards to the efficiency on batted balls, so we are able to ignore walks and strikeouts. If we restrict issues to balls in play, arm angle’s correlation to wOBAcon falls to .12, whereas its correlation to xwOBAcon rises to .26. If I needed to guess, I’d say that it has one thing to do with spray angle. When the pitcher has the platoon benefit, batters have a tendency to tug the ball much less typically, and pulled balls are inclined to outperform their anticipated stats by fairly a bit. Perhaps that variance is throwing issues off, however I can’t draw any form of straight line between that truth and the variations on this desk. Somebody smarter than I’m goes to should unpack this. As many FanGraphs readers match that description, please put any concepts you might need within the feedback. I’d love to listen to them.