Baseball’s Tower of Babel: Post 67

My dad is a preacher and I’m prone to religious references.

Baseball has a lot of technical terms. Unfortunately, many quantities are know by 2-3 names and the problem grows every time we obtain a new metric. I’ve been at this for about 2 years and I am amazed how hard it is to be sure you know the definition of a tabulated quantity.

In some cases, someone just chooses a word they feel describes something better than another, and these are pretty easily handled. An excellent example is the description of gyro angle on a pitch.

Efficiency = Active Spin.

First, here’s a quick explanation of what these refer to

Driveline, Rapsodo, Yakkertech and other influential groups prefer the former, MLB and its Savant site use the latter. The issue with Spin Efficiency is that it implies more is better, which is often but not always true. MLB acknowledges that higher efficiency is not always better, For this reason, MLB coined the term Active Spin, which you can read about here: http://tangotiger.com/index.php/site/comments/statcast-lab-active-spin-percentage…

Personally, I see inactive and inefficient as similar. The real eggheads will tell the gyro angle. Efficiency = cos(gyro angle) *100. So 0º gyro is 100% efficiency or active spin, 90º gyro is 0 efficiency. If you are computing ball movement, it’s the angle you need.

Other differences stem from a desire to make complicated information more accessible. MLB feels strongly that it is better to talk about how a ball moves than the axis it spins on. Scour Savant and you will see scant mention of “axis”. Instead, you see movement numbers directions based on a clock face. You could say “I understand what axis means, so I’ll just use the sites aimed at me.” The trouble is that MLB currently withholds some numbers from the public and they are only available on Savant. Efficiency (active spin) is one example. As a result, Savant is useful to casual fans as well as serious analysts.

So, let’s wade in movement and spin. Let me start by mentioning that Driveline has a good article on this topic. One thing I will do that they did not is reign things in a bit. This topic is easily confusing due to differences in spin axis versus spin direction and catcher vs pitcher’s view. I will delineate axis and direction, but this post will all be from the pitcher’s view. If you are interested in the batter’s view, flip it left to right.

“Back in the day,” we thought movement was all due to spin (most movement metrics remove the effect of gravity, more on that below). Our descriptors of ball spin/movement grew up around that idea. I’ve tried hard to stick a fork in that over the last year. So a lot of this is my fault.

The Statcast system uses an angle origin that is shown below from the pitcher’s point of view. When we talk about the spin axis of the ball, think if a rod through the ball on which it spins (the green line). A straight over the top fastball is therefore 180º axis. A curve is 0º by the right hand rule. The pitch shown, from a RHP with a 3/4 arm slot, is about 219º axis.

The trouble started when someone noted that spin makes a ball move 90 degrees from its axis. So “spin direction” and “tilt” were born. The white arrow shows tilt, or the direction the ball will move due to spin (i.e. Magnus effect). This assumes no other forces are at play and is shown on a clock face, which many find to be a natural way to look at this. This has been adopted by many entities including MLB (Savant), Rapsodo, Diamond Kinetics, and Pitchlogic.

It’s easy to see from that picture that 180 minutes (3 hrs) is 90 degrees. Note that the spin axis is never expressed in clock units.

For the pitch shown, the spin axis is about 219º, tilt is 1:18 and the spin direction is 129º. If I had it my way, we’d never speak of spin direction again. If you like that point of view, why not go with Tilt? Many don’t like Tilt, but given the large following it has, it’s not going anywhere.

OK, this brings us to 2020, the year we finally became sure that not all movement is due to Magnus effect. That deals a pretty big blow to some of these ideas and makes the landscape quite a bit more confusing.

MLB’s new in-game measurement system, HawkEye, measures directly and reports the axis of the pitches. My collaborators and I have been referring to that as an Observed Axis. It doesn’t seem to be catching on, but I heard a suggestion that it be called the True Axis, which is what it is. This is the axis reported by Rapsodo and Yakker. I am not sure, but I think it is also the axis reported by the “smart baseballs” made by Diamond Kinetics and Pitchlogic.

Trackman reports an axis of the pitch based on the way it moves, assuming Magnus-only movement. We’ve called that an Inferred Axis. This is what the analytics community was using until 2020 before HawkEye came online. Note that if you have movement data, you can compute the Inferred Axis, so we can still get it from HawkEye data.

Driveline, in their article on the topic, used the terms Observed Tilt and Inferred Tilt. The ideas are the same, but Tilt refers to direction rather than the axis. Referring to tilt in degrees is confusing, IMO, but I’m not in any position to throw stones here having committed nearly every usage sin.

This week, MLB weighed in by adding new data to the Baseballsavant.com site, specifically a spin direction leaderboard. There is exciting new information here including information on non-Magnus movement, and more important to me since I already had that, efficiency (called active spin). Since MLB wants to avoid the term “axis,” referring instead only to terms based on direction. All of their values are presented in clock units.

The new terms are Spin Based and Observed movement. This may be because I am not the sharpest knife in the drawer, but I immediately connected Observed Movement and Observed Axis, which is not correct. In fact, getting burned here is what motivated this entire post.

Observed Movement is tilt based on where the ball goes, so it is related to Inferred Axis.

Spin Based Movement is what you would expect from a Magnus model, so it is related to the Observed Axis.

I can both see this point of view and I am very frustrated by it since this is clearly going to generate more confusion. The folks at MLB and I disagree about whether the public is ready for “axis”. I just noticed that when you flip a picture in Photoshop, it asks you which axis to rotate around. But we’ve agreed to disagree on this point.

We defined Axis Shift to be the difference between the Inferred and Observed Axis. Driveline termed this Axis Deviation, and MLB (Savant) has adopted that term. I probably will too. Note that when speaking of differences, values computed from Tilt and from Axis are the same since they are just offset by 90º. In other words, Savant’s deviations should be the same as our Shifts. I have done a bit of poking around and this seems to be true. Note that their pitches are classified independently from the data I use (from BaseballProspectus), which can lead to differences.

Ways of Computing Movement

Perhaps worse than two names for the same thing is the same name for two things. Movement presents that kind of problem.

Movement is always defined as the deviation from a straight path at release. That removes the effect of the release. I recommend this video by SimpleSabermetrics that explains this calculation. Horizontal Break (HB) is simpler since gravity is never involved, but one does have to be careful when comparing numbers that they were all computed over the same distance, since break increases like distance squared. Savant measures movement from release to plate-crossing and that others may measure it differently.

Vertical Break (VB), however, is computed two fundamentally different ways. One is to remove the effect of gravity. In this case, back spinning pitches have positive VB. This is often called Short Form VB and is one of the outputs of a Rapsodo (based on a Magnus model–not directly measured). The other is to leave gravity in, resulting in Long Form VB, which is negative on any pitch. Yakkertech uses Induced Vertical Break for this quantity with gravity removed and calls the long form simply Vertical Break.

In addition, there are two ways of computing the effect of gravity. Alan Nathan pointed out that as a pitched baseball falls, it experiences drag that impedes its vertical movement. Some calculations account for this while others do not.

Did I miss anything? Get something wrong? Misrepresent your site of technology? I’d love to hear about it and will be updating this post, probably frequently! Thanks.

Related Post

Leave a Reply

Your email address will not be published.