clock menu more-arrow no yes mobile

Filed under:

Statcast and the coming Singularity

We’re in the early stages of completely changing how we evaluate players.

St Louis Cardinals v Arizona Diamondbacks Photo by Christian Petersen/Getty Images

At some point, you’ve probably heard or read someone refer to “the singularity”. A singularity can be many things, but “the singularity“ usually refers to the idea of a Technological singularity. There are many ways to define this concept, but it’s basically the idea that humanity will inevitably create an artificial intelligence greater than itself, which will trigger exponential and unforeseen changes to the world as we know and experience it.

While the concept has been around for over a hundred years, we can more clearly see this inevitability today than before. Even if you don’t subscribe to the idea, it’s hard not to see how technology continues to reshape our lives. We live among real, living Cyborgs. Several large companies are competing to be the first to bring self-driving technology to market. Amazon’s warehouses are so autonomous that they require just a minute of human labor to ship a package, from taking it off the shelf, packaging it, and sending it to the correct mail truck. Oh, and then there’s their drone delivery ambitions.

Okay, cool, but how does this relate to baseball? The best example of technology taking over in baseball is MLB’s Statcast technology. If you read this blog, you probably like baseball enough to at least be aware of Statcast. For the uninitiated, MLB’s glossary describes Statcast as “a state-of-the-art tracking technology...capable of measuring previously unquantifiable aspects of the game...using a series of high-resolution optical cameras along with radar equipment” to track “the location and movements of the ball and every player on the field, resulting in an unparalleled amount of information covering everything from the pitcher to the batter to base-runners and defensive players.“

I have often used the Statcast data hosted at to analyze hitters. That was thanks to Statcast being able to track both the Exit Velocity (the speed at which the ball leaves the bat) and Launch Angle (the vertical angle at which the ball leaves the bat) of most batted balls. From those two basic stats, a lot can be built on top. For instance, for each combination of the two, you can find it’s average production, whether in terms of hit probability, HR%, wOBA, BABIP or any other metric you can think of.

There’s a lot of other neat things you can figure out though. For instance, Joe Trezza wrote about how the Cards’ pitching staff has worked hard on holding runners on better. Carlos Martinez, Adam Wainwright, and Mike Leake have all cut a half a foot or more off the average lead a runner takes on them. They’ve also worked on being quicker to the plate, another aspect Statcast tracks.

After soon-to-be-former Cub and noted clubhouse cancer Miguel Montero blamed his pitcher to the media after allowing 7 stolen bases to the Nationals, Travis Sawchik used Statcast to investigate. The technology tracks each catcher’s pop-time (the time from receiving the ball to releasing it on an attempted steal), as well each pitcher’s time to home.

Montero has the worst average pop-time in the majors this year at 2.12 seconds, and the average is about 2 seconds flat. A very experienced scout can certainly detect that extra tenth of a second, but he still can’t quantify it without a stopwatch, diligence, and attending several games to get a good feel for the players’ average. He’d have a good sense of average, but he’d need to keep excellent records to find out what that extra tenth of a second means in terms of throwing out runners.

There are several other Statcast stats listed in the glossary linked above, some of the more notable being Spin Rate (the speed at which the ball rotates, something our own Joe Schwartz has often used in his pitching analysis), Route Efficiency (how close to optimal a defender’s route was to the ball), and Catch Probability, which uses a ball’s hang-time and the distance a defender had to cover to get there to generate the average chance that a ball is caught.

One new feature is Sprint Speed, released this past week. The point of this stat is to find the average max-effort speed of a runner. In a truly shocking result, Billy Hamilton is leading in Sprint Speed in 2017. They also have a leaderboard, and a really cool graphic to go with it:

for a more interesting version of this picture, check out the leaderboard linked above, which has the same image but it tells you who each dot is when you hover your cursor above it.

Despite coming out just days ago, our fearless leader Craig Edwards has already tested the stat’s relationship with base-running and defensive value in the current year, and fellow Cardinal blogger Zach Gifford has already looked at the predictive powers of the stat, as well as where the Cardinals’ regulars rank.

The point is, these all are things that are done by scouts. How hard the ball comes off the bat, the average lead a runner can get away with, a catcher’s pop-time, a pitcher’s time to the plate, an outfielder’s speed and efficiency can all be assessed by observing a player. But can they see, remember, and properly aggregate every single time a player showcases those tools? Of course not. Technology is already way better at this than humans. Scouts don’t judge a pitcher’s velocity by sight. They use a radar gun. And they don’t even need to do that at the MLB level. They can just look at their Pitch F/x numbers.

At the same time, it’s upending the way we look at stats. When examining changes in a player’s contact quality, we used to look at a breakdown of his Hard%, Medium%, and Soft%, but now we have average Exit Velocity and Barrel%. We used to look at a player’s Ground ball/Line Drive/Fly ball profile, now we look at their Launch Angle distribution. We use to have Speed Score, but now we have Sprint Speed. The best public defensive metrics - Ultimate Zone Rating (UZR) and Defensive Runs Saved (DRS) - look set to be dethroned by the fruits of Catch Probability and whatever else the braintrust working on Statcast dream up.

There’s also the fact that Statcast still has more potential to spare. Remember above I mentioned that Statcast tracks the ball and every player. Noticeably absent is the bat. Statcast offers a lot of improvement in terms of measuring performance at the plate, but tracking the bat opens up another world of possibilities.

I often think of a hypothetical application, which in my head I call Bat F/x. Perhaps a more suitable name would be Batcast. Anyway, the idea is that you could gain a lot of information from tracking the bat that is currently still something only scouts can observe. Swing velocity and Swing plane are two more obvious examples. These can be measured with special bats as a method of practice and training, but I’m talking about an in-game solution that evaluates performance.

Another neat one I would want to see would be a bat heatmap. That is, a heatmap of the half of the bat facing the pitcher at the point of impact. Then it could be color-coded based on where the hitter’s bat most often came in contact with the ball.

Perhaps Statcast just isn’t advanced enough to do that yet, I don’t know the technology well enough to say. Other technology is able to though, and it has some very obvious use cases for evaluating talent. Humans have already hit a wall when it comes to what they can reasonably do to evaluate talent by eye. Technology offers endless possibilities.

All the way back in 2004, when sabermetrics was gaining steam but still wasn’t dominating front offices like they are now, Dayn Perry had this gem of a quote:

A question that's sometimes posed goes something like this: "Should you run an organization with scouts or statistics?" My answer is the same it would be if someone asked me: "Beer or tacos?" Both, you fool. Why construct an either-or scenario where none need exist? Heady organizations know they need as much good information as possible before they make critical decisions.

Statcast represents the ultimate combination of scouting and stats. A beer-flavored taco if you will. Okay, maybe the analogy breaks down there, because that sounds horrible. Statcast technology is able to scout better than the most observant and persistent scouts, through its ability to directly measure a player’s tools and properly aggregate them over time. At the same time, it’s going to upend the existing set of stats we used before. Statcast offers a brave new world of player evaluation, and I for one am going to enjoy seeing what comes next.