I Can't Stop Talking About Felix's Fastball

One of my many impossible dreams is figuring out how to accurately separate Felix's two-seamer from his four-seamer in the PITCHf/x data. I call it a dream because that would be awesome, but I call it impossible because there's just no way to do it and be confident about the results. Where you'd like to be able to sort the numbers by movement and come up with two distinct pitches, with Felix it's basically just a random scatterplot, with lots of presumed two-seamers, lots of presumed four-seamers, and lots of stuff in between. And that's messy. Sometimes I wonder if Felix even knows which one he's throwing when the ball comes out of his hand.

The impossible nature of the problem doesn't stop me from trying to solve it, though, and while I have yet to reach or even get within sight of my destination, sometimes I do end up with neat bits of information I wouldn't have thought to look for were it not for playing around with giant spreadsheets. Contained herein are some of those neat bits, as today's project turned into separating Felix's fastball(s) by velocity to see what comes out. Because why not?

This graph should be self-explanatory:

Felix's average fastball this year clocked in at 94.8 miles per hour, and 83% of them were between 93-97. I decided to split them up into the seven groups shown above. They're not equal in size, obviously, but that doesn't matter. I just wanted to see how the numbers broke down.

Here's the fun stuff:

As expected, the faster the fastball, the lesser the movement. That's what happens when a pitch spends less time in the air. The really interesting data, though, is over there at the right. For one thing, Felix had the best control of his fastball in his comfort zones, and lost a little something at either velocity extreme. He threw 61% strikes at 96+, 66% strikes between 93-96, and 60% strikes below 93. When he either tried to amp things up or make the ball dance, his location became more spotty.

But look under that last column. The swinging strikes tell a bigger story. Felix has always drawn criticism for not being able to turn his velocity into more missed bats, but maybe that's too simplistic. Based on the data we have from 2008, it seems like Felix did turn his velocity into more missed bats - just only at his highest speeds. You're talking about a swinging strike rate of 7.8% above 95mph and just 4.0% below. Not that 7.8% is really all that terrific, but it's double the rate he had on his slower fastballs. It looks like velocity may play a bigger role than movement in determining whether or not a batter swings through a Felix heater.

However, Felix's slower fastballs did have the benefit of finding the zone more often (64.7% strikes to 63.3% strikes) and generating more groundballs (52.4% to 46.4%), so it's not like his 91-95 stuff was a pile of crap. It was just eminently hittable. I'm assuming that Felix dials it up when he's looking for a strikeout and slows things down a little bit when he wants a grounder. Of course, he has three other pitches that miss more bats and his changeup led to more grounders than anything else, but whatever, that's a different post. This is just about Felix's fastball.

One last interesting thing I saw in the data deals with final pitch location:

p_z is the measurement of pitch height as it crosses the front of home plate. A p_z of 2.0 means the pitch was two feet off the ground, and a p_z of 0 means the pitch was in the dirt. Here we see a steady trend where, as Felix's fastball got faster, it wound up lower in the zone. That's the opposite of what I expected. Ordinarily you see really hard fastballs thrown in the upper half, and sinking, runny fastballs somewhere near the bottom. Felix did things the other way around. I don't know what this means - if anything - but it strikes me as more than a little weird that Jarrod Washburn seems to trust his high fastball more than Felix does. I guess you just do what you have to do to get by when you basically only throw one pitch.

By the way, if you feel like playing around with a bunch of numbers in a spreadsheet, here are all of Felix's pitches from 2008 that were tracked and measured by PITCHf/x.