Deciphering the Braves’ Super Secret Pitching Formula
Last week, MLB.com’s David Adler wrote up an insightful comparison between Shane McClanahan and Spencer Strider. It details how both of them release their fastballs far closer to home plate than the average pitcher, allowing them to deceive and strike out batters. What’s truly interesting is some of the things Strider said about himself, or rather elected not to say:
“There is an ideal number, technically, within an equation, that diminishes to a certain degree the hitter’s ability to decide when to swing and where. But that’s a formula that I can’t control necessarily. So I kind of have to take what [my extension] usually is and then alter what I can control to optimize it.”
Strider is a cool guy, for obvious reasons — he’s the de facto ace of the Atlanta Braves, he’s nicknamed ‘Quadzilla,’ and he rocks a mustache like no other — but an underappreciated aspect about him is just how nerdy he is. He basically narrates the entire part of the article about him, and it’s quite obvious that he’s very knowledgeable about modern pitching analytics.
Midway through the article, Strider casually mentions this… formula? that combines “fastball spin metrics, your velocity, your extension, your release height” to calculate how deceiving you are to the opposing hitter. It’s implied that Strider regularly uses this formula to measure how his fastball is performing beyond publicly known measures like vertical movement. And as we can see in the quote above, there’s apparently a diminishing return on this formula, which you can imagine looks like some kind of parabola, meaning that adding more spin or increasing your extension is only beneficial up to a certain extent.
As a baseball writer who pretty much focuses on the numeric side of things, this is like striking gold. The barrier between public and team research is as ironclad as ever, and getting even the smallest insight into how the sausage is made on the other side is difficult. So Strider dropping multiple sentences — paragraphs, even — on the proprietary formulas he uses is a godsend.
The formula’s inputs are nothing new; you can find every one of them on Baseball Savant, and even something like vertical approach angle becomes easy to calculate once you know how it works. The real intrigue here is its output: The Stuff+ and Location+ models you see out there use results (groundballs, strikeouts) or the aggregates of those results (run values) to determine whether a certain pitch is valuable. But according to Strider, the Braves’ formula measures how much a hitter is expected to be deceived. This is something that is much, much harder to work out. What does one even mean by ‘deception’? And how do you translate something as abstract and mysterious as deception into a single number or percentage?
The simplest way to go about defining deception is, well, by proxy. Every model basically has three components: an output variable, an input variable, and an error term. The error term consists of random variation, but more importantly, it can also consist of the variance that the model failed to capture. So if you have a model that derives, say, the ‘expected’ run value of a pitch, then compare the expected run value to the actual run value, you can attribute the difference between the two to the existence of deception. You can plausibly infer that the reason why a certain pitcher overperforms his expected run value is that he is deceptive: The pitcher hides the ball well or throws it in a manner that the model’s inputs can not pick up. That difference between theory and practice could be what Strider uses to evaluate his deceptiveness.
As simple as this method may be, it’s little more than an educated guess. There’s no way of actually telling whether the gap between an expected and actual run value of a pitch can be traced back to a pitcher’s deceptiveness. A better, albeit more complex way to go about defining deception is by constructing a model that directly tries to quantify deception. Easier said than done! But in all likelihood, this is probably what the Braves are trying to do. Strider gives us this clue to build on:
“Not to say I’m in a sweet spot, but I think I’m currently where I’d like to be, where I’m reaching an extension that is indicative of healthy mechanics, and then the ball is spinning in the correct range for that extension,” Strider said. “But it’s coupled with a low release height as well, the [vertical approach angle], hiding the ball — and there are other things that aren’t necessarily measurable that go into whether high extension is useful.”
There are other things that aren’t necessarily measurable that go into whether high extension is useful! Now, I don’t know if Strider means these things can’t be measured, period, or if they can’t be measured using the technology and data available to the public. If I had to guess, it’s likely the latter — Strider seemed careful in choosing what to reveal — and I’d also guess that those-things-which-are-not-measurable are related to biomechanical data. Tracking the human body and converting its exertions into usable information is the next frontier in baseball, and it happens to be highly relevant to deception. Want to know how long a pitcher hides the ball before releasing it? The only way to find out is by timing the duration the ball is obscured by the pitcher’s delivery.
Earlier, Strider mentioned that the formula isn’t something he can “control necessarily,” which fits into the idea that there are biomechanical data involved. Relatively speaking, it’s easier for a pitcher to adjust his release point or extension than it is to add torque to his delivery (though I could be wrong on this, in which case, please let me know). It’s impossible for a pitcher to add an inch of height or lengthen his arms (unless we’re in 2078, and growth hormones have been legalized by Rob Manfred the Third). Those are some of the constraints that (a) a pitcher has to learn to work with, and (b) are extremely important, at least from what we can tell, in quantifying deception. Strider isn’t tall, and his limbs aren’t long by starting pitcher standards. But he has a strong lower half that drives his body forward to maximize his extension — and ultimately his degree of deception.
But I think what’s really sweet about this super secret Atlanta Braves deception-quantifying formula is that helps inform a pitcher what type of fastball he should throw to succeed. Again, from Strider:
“[My extension] shortens the distance between the ball at release and the hitter, theoretically giving him less time to decide,” Strider said. “And if I can couple that with backspin, creating the rising effect, it creates that environment for the brain — the hitter’s brain — to not perceive exactly where the ball’s gonna be when it’s time to swing.
In Strider’s case, it was the other way around: He first developed a hard, riding fastball, then worked on strengthening his lower half during rehab from Tommy John surgery to complement it. Either way, you can imagine how such a formula can unlock new insights. It’s common knowledge among pitching aficionados that a fastball with high vertical movement and a low release point works best when thrown closer to home plate. But it’s less common for someone to know that a fastball with low vertical movement and a high release point works best when thrown further from home plate:
“In the same way, a sinkerballer, for the most part, would like to increase the distance to give as much room as possible for the ball to move, for the same purpose — it diminishes the hitter’s ability to pinpoint exactly where the ball’s gonna be.”
When Josiah Gray first came up to the big leagues, he threw a fastball with similar characteristics to that of Strider — less velocity, for sure, but relative to that velocity, Gray’s own heater generated great amounts of movement. He seemed set for success. But in his first two seasons, hitters absolutely pummeled the fastball. This wasn’t just due to a lack of command; even when Gray located his fastball within the zone or parts of the zone that were ideal, hitters still found a way to make loud contact. Gray, from a low release point just like Strider’s, threw an objectively good fastball, yet could not figure out how to get major league hitters out.
What he lacked was Strider’s elite extension. Gray’s extension was, and still is, several inches behind that of Strider, which gave hitters ample time to react to the fastball. In real-time, it’s a difference of milliseconds, but those are all professional hitters sometimes need. This probably isn’t entirely why Gray struggled in his rookie and sophomore season, but it’s an explanation that makes a lot of sense. From early on in his career, Strider learned how to optimize his delivery to match the shape of his fastball; Gray, on the other hand, may not have known that his delivery was suboptimal until recently.
This season, Gray has undergone a couple of transformations. He’s no longer throwing the four-seamer as often as he used to, and its shape has been altered to a more ‘neutral’ one, featuring near-equal amounts of vertical and horizontal break. At the forefront of his repertoire is a cutter, which theoretically benefits from the extended travel time from release to home plate. So far, the results have been solid: Gray has a 3.00 ERA in 13 starts, mainly because he’s allowing far fewer home runs — in other words, hard contact — than he did in the past. There’s no guarantee these sunny days will last, but what matters is that Gray seems to have recognized his shortcomings and turned his weakness into a strength. A poor extension isn’t a problem when you throw the right kind of fastballs.
We’ll never know what goes into the Braves’ super secret pitching formula or similar ones used by other teams, but their value is very, very clear: Without them, guys like Gray or Strider may have never enjoyed the success they do now.
