Failure Files: Far From AverageBen Clemenson November 11, 2020 at 2:00 pm

Here’s the honest truth about baseball analysis: Most of the ideas I look into don’t work. That’s mostly hidden under the surface, because it’s not very interesting to read an article about absence of evidence. Hey, did you know that batters who hit very long home runs see no meaningful effect on the rest of their performance that day? I did, because I looked into that at one point, but imagine an article about that and you can kind of see the problem. Read a whole thing looking for a conclusion and find none, and you might be more than a little irritated.

Now that I’ve told you how bad of an idea it is to write about failed ideas, I’d like to introduce you to an article series about ideas that didn’t pan out. I know, I know: I was bemoaning the difficulty of writing such an article just sentences ago. Some failures, however, are more interesting than others, and I’d like to think that I know how to tell the difference. In this intermittent and haphazardly scheduled series, I’ll write about busted ideas that taught me something interesting in their failure, or that simply examine parts of the game that might otherwise escape notice.

In September of this year, I came up with an idea that spent the next month worming its way into my brain. We think of pitch movement as relative to zero, but that’s obviously not true. Sinkers rise more than a spin-less pitch thrown on the same trajectory would; they’re “risers”, in fact. Don’t tell a player that, though, because they’re not comparing these pitches to some meaningless theoretical pitch that no one throws. They’re comparing them to other fastballs, four-seamers to be specific, and if your brain is used to seeing four-seamers, sinkers do indeed sink.

Why, then, the tyranny of zero? It’s meaningless. A pitch isn’t inherently impressive because it rises 10 inches relative to gravity or has six inches of horizontal break. Movement is only useful inasmuch as it makes batters miss the ball. That’s not a secret, but we still talk about movement relative to zero instead of average. There’s no reason we have to, though.

With that in mind, I came up with a tenuous theory. Imagine the most generic possible four-seam fastball in the majors, the mean movement that a batter sees over and over again throughout their career. Over time, human brains are powerful things; they can pick up on patterns and extrapolate them into the future. When that generic four-seam “rises,” it doesn’t look that way to the batter; that’s just the arc that fastballs move in.

Here’s the theory: What if the most effective fastball isn’t the one that has the most movement, but the one that has the most abnormal movement? A pitch that has fastball spin but rises three inches less than average might be just as hard to hit as one that rises three inches more than average. When we look at a pitch and see unimpressive break, maybe that lack of movement is in itself impressive, as long as it’s keeping far away from average.

To test this, I worked out the average four-seam movement and calculated how far each four-seam fastball thrown in 2020 was from that point, per Statcast movement data. If you can imagine a series of concentric rings around that central point (3.5 inches of arm-side run, 15.8 inches of rise), we can divide each pitch into one of those regions. Here’s the breakdown of pitches by zone, as well as how hard those pitches were thrown on average:

There doesn’t appear to be much pattern to the velocity aside from the fact that the biggest misses — a fastball that breaks more than a foot differently than averag, often either a mis-classified pitch or a mis-throw — are slower on average, which makes sense. Is there a clear pattern where pitches that come closest to “average” are easier to hit?

Technically speaking, you could argue that there’s an effect here: Pitches thrown with between three and eight inches of deviation from average break have a statistically significantly higher whiff rate. The effect is tiny though, to the point where controlling for the extra 0.3 mph of velocity that bucket has on average makes the effect vanish, and there are plenty of other confounding variables as well. Even if it were there, it’s something like one extra whiff per 100 swings — hardly the effect I was hoping to find.

I started doing some variations. Some of these pitches are bound to be mis-classified sinkers, which means that a pitch that I list as having big separation from average might merely be an average sinker. To control for that, I also calculated average sinker movement. For each four-seam fastball, I then calculated the distance from each average pitch movement and took the smaller of the two numbers. The average sinker’s movement is seven inches different from the average four-seamer, with more arm-side run and less rise.

Next, I used the same bucketing technique again. You can’t picture this as a ring of concentric circles anymore, but here’s the data in tabular form:

Again, there’s not much to see here. If you’re looking for some grand effect based on movement differential, you’ll need to keep looking. The pitches with dead average movement do indeed look a little worse, but not by enough to be meaningful. It’s another one-whiff-per-100-swings kind of deal, unfortunately.

I had one more idea: what if classifying these pitches based on distance from the center is the wrong way to go about it? Vertical break is intuitively harder to hit than horizontal break; the bat goes through the strike zone horizontally, after all. Let’s re-classify the pitches based on how their vertical break differs from the average four-seamer:

It appears that direction does matter, and far more than some mythical “deviation from average.” Pitches with less vertical break than average are easier to hit, period. There’s no effect where being three inches lower than average fools batters more than being dead on average, either: the less vertical break, the more hittable, except for the pile of misclassified pitches in the lowest bucket. The answer wasn’t my pretty theory. Instead, it was just what you’d expect without thinking about it too much: The pitches that move the most miss the most bats.

This isn’t the first time I’ve come up empty on a theory about baseball, and it won’t be the last. I thought it was instructive, however, in the perils of trying to get too fancy. My initial theory had the benefit of sounding impressive: concentric circles, and something about the brain seeing patterns. It was enough to make me feel smart, like I’d uncovered a hidden truth. The answer was far more mundane: more rise means more whiffs.

Fancy theories are fun! They aren’t always right, though, and knowing when to give up on your pet theories is a useful skill. I wanted this one to be right so badly that I ignored the obvious answer, and the obvious answer was right all along. There’s no shame in learning that, even if I took a silly and needlessly complex route to it.
Read More