Why I Don't Like Motion Capture

Short Answer

I don't believe the motion capture databases (we're talking the motion capture in a lab with the dots all over the pitcher) contain the proper information. Modeling the data based on averages, even the averages of the "elite" still falls short, because the truly elite aren't doing (publicly at least) motion capture bullpens. And if they are, I highly doubt they're throwing them anywhere near game intensity. I'd love to be proven wrong on this, because it would mean there is better information in the world than I thought, but until then I'm a skeptic.

Long and Winding Answer: The Recipe for Success

The recipe to becoming the best pitcher in baseball recommends you do 3 things:

• Throw 100 mph
• Throw a lot of strikes
• Throw 100 pitches per game for ~30 games a year.

It's absurdly difficult to do even one of these things at an MLB level. But doing all three makes you a unicorn. This is because the mechanical strategies a pitcher employs to accomplish one of these tasks are usually at odds with the strategies required to accomplish another. Human bodies can accomplish the same tasks through a myriad of different ways, but the more specific the task, the fewer options. And becoming the best pitcher in baseball is hyper specific.

Narrowing Options

There's 1000 different ways to throw 95mph. But some of those ways are flawed and limit you from throwing 96mph. To get to 96mph you can only choose from 500 different ways. But of course some of those ways are flawed as well and you'll never be able to climb to 97mph by using them. There's only 250 ways to throw 97mph, but of those 250 many of them are inefficient and won't enable you to reach 98mph. Of course if you want to throw 99mph there's only 100 options available to you. And to throw 100mph you're picking from 10 different choices.

But velocity is only one ingredient for becoming the best pitcher in the game. A pitcher needs to throw strikes too! Combining velocity and control cuts the available options by at least two thirds. What was 1000 different ways to throw 95mph, is really only 333 ways if the pitcher wants to throw strikes while doing it. Which means at 100mph there's really only 3 different ways of doing it while consistently throwing the ball in the zone.

The final requirement is health. And the same logic applies. Cut all the ways you can move in half once again, and you've reached a good theoretical idea of how restrictive the strategies for movement get at elite levels. Of the 1000 ways to throw 95mph, only 167 of them can be done healthily and for strikes. And if you want to throw 100mph? 1.5 ways of doing it. Of course, this is a concept not exact science, but for evidence try syncing up the throws of guys throughout the years that healthily throw 100mph in the strike zone. They look extremely similar and this isn't an accident. They've all found the same 1.5 ways of moving to accomplish this hyper specific task.

Who's in the Motion Capture Databases

For training available to the public, it's going to be a majority of high school and college players, with a small percentage being minor league pitchers. A tiny, less than 1% fraction of the pool will be guys with major league experience. And of the MLB guys, most are likely doing publicly available training because they're struggling and looking for answers, not because they're the pitchers dominating MLB. And finally, even if you did have one of the unicorn MLB starting pitchers that throws upper 90's throwing a motion capture bullpen, the likelihood this pitcher is throwing anywhere near game velocity is ~0%.

So what we do have are large compendiums of all the ways a pitcher can throw suboptimally. A hundred thousand data points on the non ideal ways to do it. And likely 0 data points on the best way to do it.

Some Information is Better Than None

Yes, but not in the way you might think. As mentioned earlier, a lot of the common strategies for throwing fast are at odds with the strategies for throwing strikes. And those are at odds with throwing healthily. So data of a pitcher throwing 95mph that doesn't account for his ability to stay healthy or throw strikes, is statistically unlikely to be good data. There's a much greater chance the pitcher throws 95mph without the ability to healthily throw strikes than with. And that's pretty useless.

But we do have lots of data on the best pitchers that can do it all. It's on video every time they pitch, just not captured in a motion capture lab. There's a growing number of programs that can map the body from a video screen only, and they have the benefit of being able to cherry pick only the best pitchers in game scenarios for their data set. And if you don't have that information, your own eyes with slow motion video will have to suffice. Indeed, some information is better than none, but filtering by pitcher skill is more important than whatever enhanced data is gained through motion capture of the unskilled.