Confessions of a baseball analyst

© Steven Branscombe-USA TODAY Sports

Jack Leiter will always have a special place in my heart. The Rangers’ best pitcher was the subject of the very first article I wrote for FanGraphs, which talked about, among other things, his incredible fastball carry and how it could lead him to success in the big leagues. But we haven’t checked on Leiter in a while, and well, his Double-A numbers have been abysmal: a 6.24 ERA in 53.1 innings pitched somewhat dampened the hype surrounding the right-hander. While that doesn’t really change our view of Leiter, it’s still disturbing to see.

Part of that has been his inability to throw strikes, as Leiter throws out well over five walks per nine innings. But more importantly, Leiter lost a significant portion of his signature fastball turn in pro pitch. Statcast data was available for this year’s Futures Game, in which Leiter’s dozen fastballs 16.1 inch average vertical break – a far cry from the 19.9 inches I calculated in that first article using TrackMan data. It might be a small sample quirk, and yet the general industry consensus is that Leiter’s fastball is no longer transcendent. It’s a real problem.

What could be the reason? Perhaps Vanderbilt’s TrackMan device was not properly calibrated (as suggested by Mason McRae), leading to inaccurate readings. But if it’s true (and maybe it’s not), how could we verify it? What I found: Using velocity, rotational velocity, and rotational axis data from the 2021 NCAA Division I baseball season, I built a model that estimates the vertical rupture of four-seam fastballs from right-handed pitchers. Once done, I grouped the data by the pitching team and looked at which schools over- or under-valued the model. Those with the largest residuals, in theory, are prime suspects of having miscalibrated TrackMan devices.

We have evidence here. Among schools with at least 2,000 straight fastballs in the database, Vanderbilt ranks ninth out of 48 in the average difference between actual and expected vertical break. As for Leiter himself? On a not-so-small sample of 721 heaters, it generated 2.5 inches more ride than expected, putting it squarely outside the confidence interval. Leiter may also not be throwing his fastball like he used to, but it looks like TrackMan’s data has helped to soften his stat profile.

Even with reduced driving, Leiter’s fastball is still ground plus, and overall he’s still one hell of a pitching prospect. But even in an age of sophisticated data, inaccuracies can be surprisingly common. TrackMan devices are operated and maintained by humans, after all, and to err is human. While having the required data is still incredibly useful, a healthy dose of skepticism – and subsequent tweaking, such as removing outliers – goes a long way to getting the most out of it.

***

Making sure we’re not misled by the data is one thing. Deciding how to represent and communicate it is another. I’ve written a lot about pitching lately, and a few of the comments have expressed some confusion as to how the pitching motion is indicated. As if baseball wasn’t complicated enough, there is indeed more than one way to accomplish a seemingly simple task.

Because life is short and precious, here are the Cliff Notes. My preference is for what is called “short form” motion, or the expression of pitch motion relative to pitch with no spin-induced motion. Fastballs “rise” relative to this designated point of origin, while breaking balls fall instead. The short-form motion mirrors how batters actually perceive pitching, as the illusion of rising is what prompts them to swing under high-spinning radiators. It also creates a clear distinction between pitch types and their behavior, preventing us from confusing changes with sliders, for example. The abbreviated movement is what you will see on baseball flyers (including Brooks Baseball) and this site itself.

Then there’s the “long-form” movement, which mirrors how heights move in real life. Fastballs still drop, but much less compared to breaking balls. That’s what you’ll find at Baseball Savant. I guess people are confused because popular sites use different methods to represent pitch movement, which is incomprehensible. But wait, there are even two types of abbreviated movement! The first, which comes courtesy of PITCHf/x, is measured 40 feet from home plate. The second, which comes courtesy of Statcast, is based on the entire flight path: 60.5 feet, minus the launcher extension. They are functionally identical, but one produces higher move counts than the other. Specifically, it hurts our heads.

Life would be so much easier if we could all agree on one metric, but given the sport we’ve chosen to ardently follow, how can we not be pedantic about baseball? – that probably won’t happen anytime soon. It’s not just the motion of the pitch that’s drowning in semantics: baseball’s trendiest pitch breaking is widely known as a “sweeper“, but in Yankee-land it is better known as”whirling.” Spinning efficiency (Rapsodo) is active spin (baseball scholar), but some analysts take offense to the first, which implies that the higher the efficiency, the better. Meanwhile, direction of rotation and axis of rotation are two entirely different things, but that’s little explained, so even clever writers will end up using them interchangeably.

Of course, I am also part of the problem. On occasion, I switch between short and long-term moves depending on what’s most convenient, in addition to omitting explanations that I assume are just not necessary. The truth is, there could be thousands of fans out there who aren’t as knowledgeable about baseball analysis as you think. It is therefore our responsibility to ensure that they are taken into account.

***

As a FanGraphs contributor, there is some pressure to get it right, given the reputation of the site and the amount of traffic. It doesn’t come to mind like it used to, thankfully, but it’s still there in the back of my mind. Not that it’s a major problem – if you care about what you’re doing, I think it’s inevitable to be at least a little ashamed of a noticeable mistake.

But you learn not to let those moments take over you. You also learn that they present great opportunities for improvement as a writer and analyst. Earlier this month, I wrote about this season’s most and least consistent hitters, determined by a series of calculations that I’ve sufficiently explained and substantiated…or so I thought. Much to my dismay, someone in the comments pointed out that I had failed to normalize batters’ standard deviations in wRC+ to their average wRC+. Failure to do so created a positive relationship between the two variables, from which many of the conclusions in the article were drawn. Ouch.

Upon review, I realized that, yes, I had made a pretty big mistake. There’s no point in starting over with a new article, but I can catch up here. First off, below are the most consistent hitters, to date, by normalized standard deviation in wRC+ (it’s the regular standard deviation divided by the wRC+ average, aka the coefficient of variation):

The Kings of Consistency, Revisited

Then, here are the less consistent hitters:

The Capricious Group, Revisited

There is some overlap: Alonso, Flores, and Wisdom are still in the top three for consistency, and Miller remains mysteriously mercurial. Based on the number of consistent last time hitters remained, a big part of where normalization played a part is in distinguishing actual streaks from mere variance. Indeed, you’ll see that the most inconsistent list is no longer a list of the greatest hitters, which in retrospect didn’t make much sense.

Still, the adjusted standard deviation has a moderate correlation with the overall wRC+, suggesting that good hitters do tend to produce streaks of brilliance. What Alonso and Co. accomplish remains special, albeit to a lesser degree. The correlation between standard deviation and shrinkage rate is no longer non-existent, but it is weak enough that it is not worth discussing. Case in point: Wisdom and Duvall, who occupy opposite ends of the consistency spectrum, are number one and third, respectively, in terms of withdrawal rate.

The takeaways aren’t radically different, but the names certainly are. I’m disappointed that I wasn’t more careful about how I presented the data before filing the article, but what’s done is done, and there’s this little follow-up to remedy what was wrong not. Although it would have been easier to ignore it altogether, I owe it to anyone who reads my work to be honest and introspective. After all, no one wants to follow an analyst who pretends to be right all the time.