What Do Hitters See, and When Do They See It?

One morning, about two weeks ago, a YouTube video made me feel like I was asleep at the wheel. Ethan Moore, a former analyst with the Rockies and the Reds, had posted a video titled “Every Baseball Analyst is Missing Something Important.” I’d like to consider myself a baseball analyst, and it sounded like I might be missing something important. And so I clicked to see what that might be.

Over the span of 36 minutes, Ethan broke down a total of three pitches, all of which were thrown by Nolan McLean to Pete Crow-Armstrong in the second inning of a late September game between the Mets and Cubs. He used this plate appearance to illustrate his central claim: There is so much happening in the handful of milliseconds between the release of the ball and the swing of a bat, and that the psychology of the hitter — conscious thoughts, subconscious expectations, muscle memories — dictates the decision of when to swing, and where, and how hard. As Ethan put it, “When the ball is in the air, on the way to the plate, what is actually happening in the mind and the subconscious brain of the hitter?”

It reminded me of another video I saw a bit earlier in the spring. It featured Vinnie Pasquantino, before he captained Team Italy in the WBC, wearing a microphone during a live batting practice session against his Royals teammate Steven Zobac. It’s meant to be a short and funny clip, and it is both of those things, but I just kept thinking about Pasquantino’s subconscious.

A running bit in the video is that, after a pitch is thrown, Pasquantino leans back and asks the catcher what pitch Zobac threw. “That was a fastball?” he asks. “Cutter,” the catcher responds. Later on, he sounds more sure he’s got it. “Slider?” “No, cutter again.”

Pasquantino hit 32 home runs and struck out 15.7% of the time last year. This suggests that his internal computer can identify in the split of a second what pitch is coming, and where to put his swing. And yet! In this video, he is wildly misjudging pitch types. Maybe it’s a Zobac-specific thing. If his spring pitch data is any indication, he doesn’t even throw a cutter, though it’s possible that he calls the pitch a cutter, even if it looks like a slider and Savant classifies it as such. Maybe the catcher himself is confused about the pitch types. (But also, isn’t the catcher calling the pitches?) But if it’s the case that Pasquantino can’t tell the difference between two pitches, how is he as successful as he’s been?

A new paper perhaps gives some clues. In early March, three authors from the Vision and Image Processing Lab at the University of Waterloo published a paper with the eye-catching title of “Interpretable Pre-Release Baseball Pitch Type Anticipation from Broadcast 3D Kinematics.” Limiting themselves to only pre-ball flight information — in other words, looking only at the movement of the pitcher’s body prior to ball release — the authors attempted to predict the ensuing pitch type for over 100,000 pitches. They were correct over 80% of the time. In more than four out of five cases, they were able to tell what pitch would be thrown while knowing nothing about the actual pitch.

When considering the implications of a study like this, the first thought might go toward pitch tipping, and whether R&D departments might be able to leverage a similar methodology to better prepare their hitters to pick up tells. But there are also the questions it raises about swing decision psychology. When somebody hangs a slider, is Pasquantino having a verbal thought in his brain that goes something like, “That’s a slider,” or does the home run swing start sometime before? As the pitcher proceeds through the windup, and drops his trunk or bends his wrist, does Pasquantino then makes a subsequent adjustment in his setup, in the position of his hands, or in the conviction of his swing?

The next level of analysis might involve answering questions like these: Why do hitters miss, exactly? Where, specifically, do they miss? Over or under the pitch? Early? Late? In a Saberseminar presentation (and a subsequent article for Baseball Prospectus), Stephen Sutton-Brown did some really interesting work along these lines. Using Savant’s new swing tilt data, he found that the swing can be broken down into two phases: The initial 100 milliseconds, when the batter must decide whether to swing using only a two-dimensional version of the pitch, and the remaining handful of milliseconds, where the shape and angle of the swing takes hold. I tried to use some of this new swing tracking information to demystify changeups in an article from last year titled “Changeups Are Weird.” Applying a similar logic as Stephen did in his piece, it’s possible to calculate something like “implied miss distance” for any given pitch, or at least it’s possible for Alex Chamberlain, and then those implied miss distances can be aggregated to see the patterns for a given pitcher.

When Savant — at some point, perhaps not this season, but hopefully next season — starts to provide information about the position of the bat in space, these miss distances will no longer be implied. On any given swing, it will be possible to understand the specific expectation of the batter relative to the pitch that was actually delivered. Same goes with contact points: It will be possible to know how much a hitter was jammed, or how close their contact came to the end of the bat. Hopefully soon, the data will be available to tell a detailed story of the outcome of every swing.

For now, I will keep wondering about what we can’t yet know. What are hitters seeing, exactly? When do they see it? How do hitters know what’s coming, and when do they know it? What are the inputs? How conscious is the process? Can they tell what pitch is coming? When is the decision made? Is the science of hitting always going to be a mystery?