clock menu more-arrow no yes mobile

Filed under:

Felix Hernandez & A New Approach To Studying Release Point

I love a lot of things. Talking about Felix Hernandez and analyzing pitching mechanics are two of them. Now, combining two things you love into one isn't guaranteed to produce something you'll love even more (like, say, a peanut butter and golf ball sandwich), but I think this might be one of the cases where it works out pretty well.

People who've read this site for, well, about a day know that every so often, I'll take some screenshots of pitchers from footage and break them down if I think I'll find something interesting. It's a pretty good way to spot mechanical flaws, but one of its shortcomings is that it's difficult to study release point consistency, because to do so you'd have to look at a group of images from different pitches and try to eyeball whether or not the ball is being released in the same place. It's a tedious, inexact procedure that isn't really worth the effort.

Well, yesterday I was feeling a little adventurous. Maybe it was the high I got from posting the Burnett piece, since talking about delivery always makes me feel all warm and fuzzy, but I thought to myself, "to hell with eating, I'm going to figure out a way to study release points." After a little brainstorming, I came up with my procedure.

Take this screengrab of Felix, just prior to release:

The ball's a little fuzzy, but you can still see it - it's definitely there, in his hand, about to fly forward towards the plate. Now here's the cool part: copy that picture into MS Paint and move your cursor over the center of the ball in the picture. On the bottom of the screen, towards the right, you'll get a set of coordinates - for me, it reads 472,325. Think of this like a set of coordinates on any generic x,y plot. The center of the ball is 472 x units (pixels?) from the left, and 325 y units from the top.

The plan, then: get a bunch of screenshots of Felix just prior to release, find those coordinates for the center of the ball in each picture, then put them into a spreadsheet to analyze for variation. Ideally, I'd be able to end up with a few neat little 95% confidence intervals for Felix's release point.

(I realize that reading the methodology is less than invigorating, but bear with me.)

Note that:

  1. I analyzed Felix's release point for 9 fastballs and 10 curveballs separately first, then as a group.
  2. I only chose images from one game (the last of the season, vs. Oakland) so that I wouldn't have to account for significant differences in center field camera angle.
  3. The 9 fastballs and 10 curveballs were chosen randomly (I had 10 fastball pictures, but deleted one because the image sucked, and I didn't really feel like going back and replacing it with a new one).
  4. To account for any differences in camera angle and image scale, I chose two reference points (opposite corners of home plate) and made an adjustment accordingly.
  5. Once I had 95% confidence intervals in x,y units, I converted to inches by using the fact that the diamater of a baseball is about 3 inches in real life and 11 units on the screen. Simple multiplication and division.


The red(dish) box around the ball represents the 95% release point confidence interval - based on the collected data, Felix will release 95% of his fastballs with the center of the ball somewhere within that box. The dimensions of said box: 4.9 by 2.6 inches.


Same deal - red box=confidence interval. 95% of the time, Felix will release a curveball with the center of the ball somewhere within a 4 by 2.4 inch box.


The box is smaller because the sample pool is larger - now I'm going by the total sum of 19 pitches. Based on this data, the center of the ball when Felix throws either a fastball or curveball will be released somewhere within a 3 by 1.6 inch box 95% of the time.

Now, without a baseline or league average, we can't really say for sure whether or not Felix's release points are picture perfect or wildly inconsistent. However, imagine if you will a box that's 3 inches by 1.6 inches - just going by things I have around me right now, I'd say that's roughly the size of a cell phone. With 95% of his fastballs/curveballs, Felix releases the ball somewhere within an area the size of a cell phone. That's seems incredibly consistent to me, and probably serves to explain in no small part why Felix's control was so good when he was promoted to Seattle.

Something else of note - Felix's curveball release point was a little more consistent than that of his fastball. Again, it's a limited data sample, but that seems to support the prevailing notion that the heater isn't Felix's best pitch.

Have I mentioned recently that Felix Hernandez is totally awesome?