Baseball analytics has become a staple of Major League Baseball over the last decade. The analytics movement can be seen on broadcasts, in newspaper and internet articles and, of course, at the Sabermetrics, Scouting, and the Science of Baseball conference. Jimmy Wulf gives us part four of his 2015 SaberSeminar Recap.
Welcome back, saber-fanatics, to the finale of my rapidly-declining-in-quality recap series of the 5th Annual Sabermetrics, Scouting, and the Science of Baseball Seminar. You can check out Parts One, Two and Three here, and SoSH’s Doug Storms is coming at you with his own pieces later in the week on many of the other analytics-oriented speakers. Now, let’s play with some PITCHf/x.
Saturday 9:15 A.M. -Matt Lentzner’s Pitching Peanut
It shouldn’t be surprising that the operator of the best public PITCHf/x site on the planet can reinvent how you look at PITCHf/x in fifteen minutes. Saberseminar co-organizer and brooksbaseball.net owner Dan Brooks was also the leadoff hitter for the event, showing the crowd interesting ways a simple pitch movement graph can be manipulated to understand the relationship between pitchers’ release points and the movement on their different pitches.
(Quick aside: Dan Brooks and Chuck Korb devote an inordinate amount of time to organizing and running the Saberseminar every year, bringing in speakers, coaxing donations from sponsors, and handling all the event logistics. They should be applauded for devoting so much time to this Jimmy Fund fundraiser. Seriously, stop reading, stand up and clap three times. OK… you may continue.)
On brooksbaseball.net, you can see pitches charted on a graph that shows the actual horizontal and vertical movement/break of each pitch. Here’s an example of how it looks in practice, from Clayton Kershaw’s start last Friday:
The dots are clustered because Kershaw has three basic pitches in his arsenal, and each of those clusters is centered around the vertical/horizontal break lengths typical to those pitch types. Roughly, the 4-seam fastball breaks ‘up’ and slightly to the pitcher’s throwing arm side, a slider breaks down and to the glove side, and the curveball breaks much further down to the same side.
More generally, here’s a look at where each type of pitch very roughly falls on the movement graph:
Those are some pretty big circles compared to the size of the clusters on Kershaw’s graph. That’s because there’s a ton of variance in how different pitchers throw each one of those pitches. The above chart is helpful as a rubric, but not something you can apply to a game.
What Brooks demonstrated for the audience was the “Pitching Peanut”, which takes the four-seam fastball location for an individual and uses it to predict the exact location that all the pitcher’s other pitches will fall on the graph, in much tighter circles than the generic view. Applied to the above chart, it looks like this:
With a 4-seamer in the exact center of that circle, the other lines point to where that pitcher’s other pitches will locate based on that starting point. If we were looking at a pitcher with a 4-seamer with half the amount of vertical break, we’d rotate all the arrows a little bit clockwise to compensate for the new location, and they’d all point to the new secondary pitch locations.
For clarification on the ‘Pitching Peanut’ term: if the chart had a couple more esoteric pitch types on it, and we drew an outline around the edge of all the arrow endpoints, it would form a peanut shape.
The reason this works is twofold. One, the distance (in x and y position) of any other pitch type from the 4-seam cluster location is consistent across all pitchers. Two, the movement of a 4-seamer is mostly dependent on the pitcher’s release point, with predictable impact when adjustments are made.
Fortunately, the Peanut isn’t just a parlor trick with a fancy name. Brooks outlined how it could be used as a predictive tool by a pitcher tinkering with his delivery, understanding what he needs to do to maximize the effectiveness of his pitches. As analytics continue to improve our understanding of pitch movement and delivery mechanics, we could see more pitchers “optimizing” their releases to take advantage of their natural movement and pitches. The true mound artists such as Johnny Cueto could combine adjustments to advance scouting and change his release from batter to batter to optimize his movement based on their strengths and weaknesses. I could also see this being used in reverse, for prospects not yet pitching live games under the eye of PITCHf/x. What we can get from video is the exact spatial release point of a prospect (especially with the advent of wearables, discussed by Dr. Bryan Cole in a later presentation) when throwing. Could this be used to model what a pitcher’s PITCHf/x will look like in the majors, and allow for mapping to historical data to predict success or failure?
With so many heavy hitters and weighty, interesting topics throughout the weekend, I’d be hard-pressed to make a case that Brooks’ walk through PITCHf/x was the most critical or important thing I heard at the fifth annual SaberSeminar. However, It was perhaps the most emblematic of two important themes running throughout the talks. One, the ‘Pitching Peanut’ is a communications tool; something that allows me, a math novice at best, to take the overwhelming thousands of data points within PITCHf/x and make sense of it with a clean and simple rubric. Is it meaningfully helpful? Who knows! Even if it turns out a dead end, it’s an interesting thing that advances our thinking another yard down the field. We can’t get enough of that, ever. Two, it was created by people outside the industry using nothing but public data, and the amount of data available publicly is an area of exponential growth in both the near past and near future. We no longer live in a world where there are serious pockets of resistance to synthesizing advanced analytics into baseball strategy and development. The next stage of the saber revolution will be defined by our collective ability to corral the data stream geysers gushing up all around us, find the signals in the noise, and be able to translate those things down into simple, coherent, and coachable conclusions.
Events like SaberSeminar, bringing together many different perspectives from the industry and outsiders, are invaluable for the disagreements, insights and inspirations they can trigger. I hope these recaps over the past week have brought a little bit of all three to you. Thanks for reading, and see ya’ll next year.