What is the Strike Zone?
Despite what baseball tradition may suggest, the strike zone is not a rectangle. The strike zone is constantly shifting, depending on the umpire, the batter, the catcher’s receiving skills, and even the count. Called strikes are a vital component to on-field success, leading to frequent objections about umpire accuracy and an ambiguous zone. Using Trackman data across NCAA Division I baseball, we created a model to predict the called strike probability of an individual pitch and identify the location of the true college strike zone.
Our model predicts the probability of a called strike, independent of a swing. To make this prediction, we utilized the location of the pitch, the pitch metrics, the batter’s handedness, and additional situational variables. This provided us with a called strike probability for every pitch in our dataset.
The college strike zone contains similarities with the traditional zone, but it expands beyond the edge of the plate and is slightly lower than a zone one might see on Trackman or television broadcasts. In reality, the true strike zone is an ellipse that shifts outward depending on the batter’s handedness.
Catcher Evaluation
Most catcher evaluation systems assess receiving by crediting the catcher for called strikes outside of the traditional strike zone and penalizing the catcher for balls called within the strike zone. This assumes all pitches outside of the zone are equally likely to be called strikes and does not distinguish between count, batter handedness, or other factors we know to be significant.
Objective strike zone data allows us to evaluate catchers with much greater detail. We can compare the called strike probability of each pitch with the actual call throughout a game to evaluate catcher framing and identify the locations and pitch types that catchers best receive. This allows us to compute how many strikes each catcher adds throughout the course of an at-bat, game, or entire season. One objection may be that this method is extremely dependent upon the umpire. While the umpire does have a significant impact on the strike zone in a single game, over the course of a full season the metrics stabilize and the strikes added metric reflects catcher receiving more than umpire accuracy.
The plot above shows a catcher with elite receiving skills on pitches away from each batter side and at the bottom of the zone. However, our model suggests there is room to improve on elevated pitches, specifically on the inner half. Since this plot contains data across multiple games, we know it is unlikely to be a reflection of the umpire. The strikes added are attributed solely to the catcher’s receiving.
Umpire Evaluation
As alluded to earlier, umpires have the ultimate decision when it comes to defining the zone on a nightly basis. We know that all umpires have slightly different zones. Judging umpires strictly upon accuracy is an antiquated and ill-informed method. Instead, we can use called strike probability to grade umpires in the same manner that we evaluate catchers. Over the course of a game, our model will identify the largest misses, as defined by called strike probability. This information allows umpires to evaluate their performance against the average NCAA umpire, instead of relying upon a rectangular zone that is proven to be an inaccurate representation of the strike zone.
In the previous plot, pitch #4 is outside of the traditional strike zone and would be considered a ball by most players and coaches. However, according to actual umpire tendencies across college baseball, this pitch is called a strike 73% of the time. Our model reflects this probability to add greater context when assessing umpires. In addition to post-game evaluation, we can scout umpires across the league and know their tendencies and zone sizes before the game begins. This enables our hitters to know which areas to expect strikes and allows our pitchers to know that night’s zone before the first pitch is even thrown.
Pitcher Evaluation
Additionally, we can use the true college zone to evaluate strike-throwing capabilities. How often a pitcher is in the zone could be a misleading metric if the strike zone, as traditionally defined, only contains a fraction of pitches that are actually called strikes. This led us to create an Adjusted-Zone percentage that reflects how many pitches a pitcher throws with a called strike probability greater than 50%. This is a much more realistic measure of control and credits pitchers with called strikes outside of the traditional rectangular strike zone. If a pitch is called a strike 70% of the time, a pitcher should not be penalized for throwing to that location, even if that pitch is off the plate and outside of the rectangle. Instead, we should teach pitchers to exploit this zone and force the batter to expand beyond the plate. If the batter swings, they are unlikely to make hard contact, given the location of the pitch. If they don’t swing, our data suggest that pitch is going to be called a strike more often than not.
Conclusion
The college strike zone is much wider and larger than the traditional box that most people envision. In addition to visualizing the actual strike zone, our called strike probability model can be used to evaluate catcher receiving by identifying where catchers gain and lose the most strikes. We can also use this metric to evaluate umpire performance in a more accurate and objective manner, as well as to quantify exactly how large each missed call was, relative to other college umpires. Lastly, by identifying the true strike zone, we can quantify pitcher control more accurately and understand that most pitches off the plate, regardless of batter side, are going to be called strikes.
The traditional rectangle is not an accurate representation of the college strike zone. By using called strike probability, we can objectively define the strike zone and evaluate players and umpires with much greater accuracy.
