Where Should We Measure Vertical Attack Angle?
Where Should We Measure Vertical Attack Angle?
The Answer Might Surprise You
Vertical Attack Angle (VAA) has seeped into the analytical jargon, allowing for a great proxy of what makes a fastball good - low release points and hitting the top of the zone. This is not a post about VAA, if you want a thorough run down of VAA, this article is a great read, and pretty much all you need to know:
NB: All data shown here are 4-Seam Fastballs, split by pitcher seasons, min 100 pitches, 2021-2023 inclusive and include a bug where foul tips are not considered whiffs.
At its heart, VAA is measuring the difference between the velocity towards the plate (vY) and the velocity towards the ground (vZ). Typically, VAA is computed using the Y and Z velocities at the front of the plate, denoted as vy_f and vz_f. This is then converted into an angle, however this is just a semantic difference, as it has a perfect correlation to just vz_f/vz_y (the ratio of z velocity at the front of the plate to y velocity):
VAA is mostly a proxy for 2 things, release height, and plate height:
We get R^2 correlations of 0.57 and 0.22 respectively for VAA. You’ll see that Felix Bautista, king of the “flat” fastball doesn’t rate particularly well from a VAA viewpoint since he has an over-the-top delivery.
VAA Works
Throwing from a low slot is generally good for fastballs. Throwing high in the zone will reduce your VAA. VAA combines these two concepts and does a pretty good job of predicting whiffs:
R^2 values clock in around .237 for SwStr% and .224 for whiff rates. Note that I have a small bug in my data that doesn’t include foul tips. It’s theoretically possible that this will impact results, but I doubt it.
Is Home Plate The Correct Measurement Point
I first spent some time trying various points along the flight of the ball, but each one was worse than vanilla VAA. Remember that we’re basically just measuring vz_f/vy_f, the actual angle reported doesn’t really have much meaning beyond what we apply to it. First I tried extending it to the back of home plate, and that improved our correlations to whiffs. Intrigued, I started moving it farther and farther back, and it turns out, that the optimal measurement point (for maximizing correlation to SwStr% and Whiff%) wasn’t the front of home plate.
I measured the r^2 values of both metrics, at various points behind home plate (numbers on the y axis are in feet). It looks like 13 feet behind the front of home plate gives us much stronger correlations across the board than the front of home plate.
Above is how the correlations look when we use VAA compared to a VAA computed 13 feet behind the front of the plate.
Felix Bautista and Justin Verlander
Do I think we absolutely should compute VAA at 13 feet behind the front of the plate? It does a good job of giving pitchers like Bautista and Verlander credit for their IVB:
Notice how with “stock” VAA they are major outliers, but putting the measurement point farther down the theoretical line, makes their predictions more accurate. 13 feet seems to capture some balance between release point, plate height and induced vertical break.
Concluding Thoughts
I’m not convinced 13 is the correct point. I’m also not convinced it even makes sense! My objective with this piece is that we should be asking ourselves where the correct measurement point is. I don’t think the front of home plate is the correct point. I encourage fellow researchers to come up with their own viewpoints, or show me why my idea is complete crazy!
After thinking and reading some... I would say there is a complex relationship between VAA and pitch effectiveness. That is, SwStk% = f(VAA, pitch type, pitch location, ...) -- there are other factors that affect SwStk% than just VAA. For example, this article (https://medium.com/iowabaseballmanagers/fastball-vertical-approach-angle-12e2824d245a) shows that flat VAA is best for fastballs up in the zone, while steep VAA is best for fastballs down in the zone. I'm thinking that VAA(-13 ft) is able to capture some of the dependencies embedded in these other factors. That said, my recommendation (not that you asked for it :)) would be to use the additional data (pitch type, location, etc.) in tandem with VAA to gain insight, rather than rely on the value of a non-physical quantity that is calculated from elements of VAA. In other words, VAA(-13 ft) may be a nice shortcut, but I suspect better insight could be gained by taking the longer route.
FWIW, from a human perception standpoint, I could possibly see the VAA calculated at some point a few feet *front* of home plate being more correlated to SwStk% etc., because VAA in front of home plate would seem to be the basis for a batter's swing decision.