The Apparent Tilt of Seam Shifted Wake Pitches: Post 59

Conventional pitches break because of two forces, gravity and Magnus force (the ball experiences drag also, but drag his little effect on the direction of a pitch). When a Seam Shifted Wake (SSW) occurs, the forces causing break are gravity, Magnus force and the SSW force.

This article will assume you are comfortable with the concepts of spin axis and tilt.  
If you are not, have a look at our post describing these terms. Also please note that 
I am interpreting tilt AND spin axis to be descriptions of the axis, not of the ball 
direction. I learned yesterday that this is somewhat controversial.

It is common for 3rd party sites to use Trackman 9-parameter pitch trajectory descriptions to find the tilt (or equivalently the spin direction) of a pitch. I will pick on the excellent site Brooksbaseball.com because everyone seems to agree that it is reliable. These sites use a method such as Alan Nathan’s to convert the 9P trackman data to a spin axis. As noted in that document, the method assumes that gravity and Magnus are the only forces on the ball causing the ball to modify its original trajectory.

But there are pitchers out there for whom these methods just do not work. I am going to use my friend Jared Hughes as an example. Here is his Brooks 2019 Spin Axis data. He tells me his 2019 arm slot was a bit higher than he is using now and resulted in a tilt on his 4-seam and sinker that was about 2:00 +- 5. It is important that the 2 pitches have a very similar spin axis.

The Brooks data (repeated below) show his average 2019 4-seam to have a spin axis of 235º (1:50 tilt) of and a sinker spin axis of 270º (3:00 tilt). Again, he believes these two pitches actually have a similar axis, and this is supported by the videos below.

The Rapsodo data he has shared with me during the pandemic shows tilt on his 4-seam and sinker that are about 2:10-2:20, with the sinker being a few minutes more than the 4-seam.

I used my pitch tracking MATLAB script and video like the one below (setup and code described here) to compare the trajectories of one of his recent sinkers and 4-seams, as discussed in Post 57 and 58.

Sinker with 2:12 Tilt (per Rapsodo).
4-Seam with 2:20 tilt.

The plot below shows the trajectory of these 2 pitches based on the video (symbols) and as modeled by UMBA 2.1 (solid lines) with no SSW model. The tilt of the two pitches are very similar, and so UMBA predicts a very similar trajectory. The actual pitches are quite different, one above and one below that simulation.

I can use the videos and Trip Somer’s spin visualization tool to estimate the orientations of the two pitches. I estimate the 4-S to have a front orientation of 28º and top orientation of 90º. I estimate the sinker to have a front orientation of -39º and top of 0º. Note that UMBA calls Top “Y” and and Front “Z“.

When I use the seam model, we see the predicted 4-Seam trajectory to move up while the sinker trajectory moves down, as shown below. The seams give his 4-S additional positive vertical movement and the sinker negative vertical movement. These predictions clearly match the measured trajectories better than the no-seam model. But, what happens if you try to interpret this trajectory as having only influences of Magnus and gravity? After all, that is what Brooks and other sites do.

To answer this question (which which first posed to me by Jared), I asked Andrew to add a new output to UMBA: Apparent Tilt.

To compute tilt, one first has to determine “Movement” or “Break”. From Andrew: I take the initial vertical release angle and horizontal release angle and draw a straight line to where a ball with no forces (Magnus, seam effects, or gravity) acting on it crosses the plate. This is marked in the figure below as “No Forces”. Movement is the difference between that location and the location of the plate of the pitch with the effect of gravity removed, called “0-g” in the figure.

Explanation of “Movement” using a curveball and a fastball as an example. In both cases, the effect of gravity is removed before computing Movement, so the vertical movement of a fastball is positive and a curveball is negative. In each case, movement is relative to a straight line in the release trajectory.

The angle spin axis (θ), from vertical, is calculated by  

θ = arctan(MH / MV) +180º. We have to add 180 because someone decided to make that upward instead of 0.

This can easily be converted to tilt based on figure below. 15 minutes equals 90 degrees. When this is applied to an UMBA 2.X trajectory, the result is Apparent Tilt. If a seam model is included and if the pitch has a seam shifted wake, the Apparent Tilt will not match the input tilt.

The relationship between movement and tilt (black) or spin axis (green).

My claim is that pitchers who throw SSW pitches often will have incorrect spin axis (which is directly related to tilt) on sites that use Trackman data to compute it, such as Brooks. To illustrate, in the table below, the tilt reported from Jared’s Rapsodo is compared to the apparent tilt for the same pitch simulated in UMBA. While he has raised his tilt by about 0:30 in general, the same trend in his 2019 data (above in orange) is apparent. The tilt of his sinker is over-reported by 0:47 and his 4-seamer is under-reported by a similar amount.

My claim is that any site like Brooks that computes spin axis based on movement data will have numbers similar to the UMBA apparent tilt.

One of the more interesting things that have come out of this was a story from Jared. “I have come across multiple team personnel in my career that have believed the Brooks/Trackman axis numbers. I have been told that my axis is around 270°, if not beyond it at times. This subsequently effected my slow motion video and Rapsodo analysis. When the axis showed up closer to 250° on Rapsodo and slow motion video, I believed that something was off with how I was releasing the baseball. While I looked to fix the axis, it is likely that this was the wrong approach. Instead of trying to achieve an axis beyond 270º, I should have been trying to replicate the SSW orientation.”

Thanks for valuable input from Alan Nathan, Tom Tango, and, our course, Jared Hughes.

Related Post

3 thoughts on “The Apparent Tilt of Seam Shifted Wake Pitches: Post 59

  1. If I understand correctly, the points on your trajectory graphs are the actual trajectory, as measured from the video. How accurate is that? A better way to do this kind of study is to utilize both Trackman (to get the trajectory) and Rapsodo (to get the spin axis) simultaneously. Then you can use the movement calculated from the trajectory to infer a spin axis, assuming all Magnus. Then compare with Rapsodo. I have such data and have done such an analysis. I’ll e-mail you something shortly.

  2. Now I understand the twitter discussion a bit more about Tilt and Spin Direction describing 2D axis or ball direction. In my mind they ONLY describe the 2D axis. The ball direction is calculated – and not all systems do it the same as you point out here.

    Unsurprisingly, I like the way that PitchTracker measures spin axis since it’s a direct measurement AND it’s measured right out of the hand. We have seen the spin axis change in flight, so I don’t trust calculations made late during flight.

    1. I agree. I think we all need to make that point whenever we can. Pitchtracker and Rapsodo measure the axis. I am sure Yakker does too.

      While I acknowledge the growing body of evidence that the axis may change, I don’t think we are ready to deal with that yet.

Leave a Reply

Your email address will not be published. Required fields are marked *