What does the college baseball strike zone look like?

The college strike zone is big, variable, and full of opportunities for competitive advantage.

Whether teaching a pitcher where to nibble, training a hitter how to control the zone, or showing a catcher how to steal more calls, we have to understand the geography of the strike zone. We have to know the strike zone as a league average, we have to know the strike zone for each umpire, and we need to know the limitations of our data.

Let’s start by examining the 2024 Division I (D1) college baseball season (see also this nice post by Keithan Sharp). Stadium Trackman units around the country recorded 1,360,519 pitches, of which 749,030 were called balls or strikes (and not swung at). Of these pitch calls, 87,334 (11.5%) were incorrect. By incorrect I mean that the umpire called a ball when the pitch was in the Major League Baseball (MLB) defined strike zone (‘false balls’), or that the umpire called a strike when the pitch was outside the MLB strike zone (‘false strikes’).

MLB.com says: “The official strike zone is the area over home plate from the midpoint between a batter’s shoulders and the top of the uniform pants — when the batter is in his stance and prepared to swing at a pitched ball — and a point just below the kneecap. In order to get a strike call, part of the ball must cross over part of home plate while in the aforementioned area.”

Figure 1 reveals that the two most common calls missed by Division I college baseball umpires are (1) balls low and/or away called for strikes (note the strong asymmetry between left-handed batters (LHB) and right-handed batters (RHB)), and (2) strikes in the upper part of the zone called for balls. Umpires call about three times as many false strikes as they call false balls.

Figure 1. Where do umpires miss calls? The static strike zone not accounting for batter height or stance (black polygon) is taken as x: -0.95 to +0.95, z: +1.645 to +3.355.. The heat map depicts a composite two-dimensional histogram, where white means no missed calls, red means an excess of false balls (pitches called balls that should have been strikes), and blue means an excess of false strikes. The two most common calls missed by Division I college baseball are (1) balls low and/or away called for strikes (note the strong asymmetry between left-handed batters (LHB) and right-handed batters (RHB), and (2) strikes in the upper part of the zone called for balls. Around 11.5% of all pitches called (i.e., not swung at or put in play) were missed calls. The data in this figure represent all 749,030 called pitches recorded by stadium Trackman units during the 2024 collegiate season.

As more stadiums get equipped with Trackman, leagues have started evaluating umpires based on their strike calling accuracy (or at least that is what the umpires tell us). Players and coaches have reported seeing a shrinking strike zone in response to these umpire evaluations, and that sentiment seems to be borne out in Figure 2, which shows 5.2% fewer missed calls in 2024 than in 2023.

Figure 2. These difference maps depict changing patterns in pitch calling in Division I baseball between 2023 and 2024. Negative numbers (purple) mean fewer missed calls in 2024 than in 2023. So for example, consider the upper part of the strike zone: The abundance of negative (purple) values means that fewer ‘false balls’ and ‘false strikes’ were called in that region in 2024 than in 2023. In contrast, the outside part of the zone is responsible for more missed calls in 2024 than in 2023 (green). Overall, umpires improved between 2023 and 2024, with decreases of 0.08 to 0.45 in percent false strikes and false balls, corresponding to a 5.2% decrease in missed calls.

Table 1 summarizes the comparison between 2023 and 2024 D1 baseball. The good news is that the number of games played in Trackman-equipped stadiums is increasing. The bad news is that as more programs install Trackman units, and more leagues and umpires are contributing data, I see an increase (not the decrease we expected after making Figure 2) in the average size and variability of the strike zone (Table 1).

Table 1. GamesPlayed and PitchesTracked refer to games and pitches recorded by Trackman (not all Division I games). All other columns refer to per-game statistics, where μ is the mean and σ is one standard deviation. For reference, the area of the ‘true’ strike zone (the black polygon in Figures 1 & 2) is 3.25 square-feet.

Table 1 prompts me to ask how Ivy League umpires compare to Southeastern Conference (SEC) and Major League (MLB) umpires. Table 2 shows that only about half of Ivy League games (84+playoffs) are recorded by Trackman. Compared to 2023-2024 patterns in the Ivy League, the SEC saw a larger decrease in false balls, and a small increase in false strikes (Table 2). Overall, the SEC has a 10% smaller and significantly less variable strike zone than the Ivy League. MLB has a strike zone another 21% smaller than the SEC and significantly less variable (Table 2).

Table 2. GamesPlayed and PitchesTracked refer to games and pitches recorded by Trackman (not all Division I games). All other columns refer to per-game statistics, where μ is the mean and σ is one standard deviation. For reference, the area of the static strike zone (the black polygon in Figures 1 & 2) is 3.25 square-feet. For MLB data, Static refers to the same static strike zone used for Division I data, while Custom refers to the changing vertical size and location of the Statcast strike zone (using the sz_top and sz_bot variables) in response to batter height and stance.

So far, I have evaluated where in the geography of the strike zone most missed calls occur (Figures 1 & 2), as well as frequency of missed calls in different leagues (Tables 1 & 2). I also wonder whether umpires are more likely to miss certain calls in certain counts… and the answer is yes! Figure 3 depicts the frequency of missed calls by count. False balls are most rare in 3–0 counts, and generally less common in hitter’s counts. In contrast, False strikes are most rare in 0–2 counts, and generally less common in pitcher’s counts. These results are true whether you study all of Division I baseball, the SEC, or the Ivy League (the mean rates just change, as in Table 2). As most players and coaches expect, umpires behave to prolong plate appearances, giving pitchers an extra chance if they just miss on 3–0, and giving hitters an extra chance if they freeze looking at a close 0–2 strike (Figure 3).

Figure 3. For each pitch count, across different leagues, I compute the rate with which umpires call false balls and false strikes. There is a single colormap for false balls, and a single colormap for false strikes, that can be compared directly between leagues.

Before moving on, let’s investigate one other (surprising) result from Table 2 — the effect of the custom strike zone in the MLB. Recall that Statcast records unique coordinates for the top and bottom of the strike zone for every batter, in response to his height and stance. Does the fact that I have to use a static strike zone when analyzing D1 data mean that I am overestimating the number of missed calls in college baseball? Well, in MLB, using a custom strike zone has no significant effect on the fraction of bad calls, it just reduces the number of false strikes while increasing the number of false balls (Table 2). Let’s investigate this result graphically.

Figure 4. These heat maps are constructed using the same methods as Figure 1, with two important differences. First, here I use 2024 MLB data instead of D1 college baseball data. Second, while the black rectangle depicts the same static strike zone used in Figures 1 & 2, the heat map depicts missed calls based on Statcast’s custom strike zone that moves up and down (but not side to side) based on the batter’s height and stance.

Examining the MLB strike zone in Figure 3, I would summarize as follows:
•The left-right asymmetry in false strikes on the outside part of the plate persists in the MLB, but is not as pronounced as in D1.
•On the inside part of the plate, MLB umpires call a lot of false balls, but virtually no false strikes.
•At the top and bottom of the zone, umpires call virtually no false balls, but call a lot of false strikes
•The custom Statcast zone moving up and down with batter height and stance reduces the number of these false strikes, but increases the number of false balls, at the top and bottom of the zone. This result tells me that either (a) MLB umpires are not adjusting their strike zones enough for batter height and stance, or (b) Statcast’s estimates of a batter’s strike zone are inaccurate. Either way, the lack of dynamic zone coordinates in D1 baseball does not seem to put us at an analytical disadvantage compared to MLB.

Now that we understand the shape, size, and count-dependence of the D1 strike zone, and have evaluated the impact of not accounting for dynamic vertical zone coordinates, let’s look at some examples of how we can leverage strike zone geography for competitive advantage.

Ivy League umpires have large and variable strike zones (Table 2), and every umpire has unique tendencies. For example, Mike Cerra (chosen at random as the umpire that has called the most Ivy League pitches) has a 12.97% missed call rate (Figure 4), which is slightly higher but within uncertainty of the Ivy League average (Table 2). Cerra calls false strikes further off the outside of the plate than most Ivy umpires, but his tendency to call false balls in the top of the zone (or really anywhere) is smaller than most Ivy umpires (Figure 4). Cerra has a 0.000 false ball rate on 3–0 counts, but a league average 0.038 false strike rate on 0–2 counts. And so on… Our pitchers and hitters can prepare for these umpire tendencies.

Figure 5. False ball and false strike tendencies for umpire Mike Cerra. The black polygon is the static strike zone used in Figures 1 & 2.

The geography of false balls and false strikes also can teach us about the efficacy of pitch framing. Table 3 and Figure 5 show three important results from the Ivy League:
•On average, catchers steal about 11% of balls and make them strikes, but also lose about 11% of strikes and make them balls.
•Lost strike rate varies about 12% more than stolen strike rate, but has a similar range.
• Usually, an Ivy League catcher has a different home plate umpire for every game, rendering attempts to evaluate the impact of umpire tendencies ambiguous due to small sample size. Eventually, with a few more years of data, I hope we can add an umpire-adjustment to the Framing+ statistic.

Table 3. These counting and rate statistics include only those games recorded in a stadium with a working Trackman system. StolenStrikeRate = FalseStrikes / (FalseStrikes + TrueBalls); LostStrikeRate = FalseBalls / (FalseBalls + TrueStrikes); MissedPitchRate = StolenStrikeRate — LostStrikeRate; Framing+ = (MissedPitchRate — μMissedPitchRate+ 1)*100. In this table, μMissedPitchRate is computed from qualified (>500 PitchesCaught) IvyLeague catchers. BRO = Brown; COL = Columbia; COR = Cornell; DAR = Dartmouth; HAR = Harvard; PEN = UPenn; PRI = Princeton; YAL = Yale.

Figure 5. Data from Table 3, where the color of the symbol reflects the team, and the size of the symbol depicts the number of pitches caught that were recorded by Trackman. See the caption to Table 3 for equations. In this case, we have computed Framing+ only using data from Ivy League catchers, so by definition, the average Ivy League catcher has a Framing+ of 100.

Compared to the Ivy League, average SEC catchers steal about 11% of balls and make them strikes, while losing about 9% of strikes and making them balls. The best performing SEC catcher is Ryder Helfrick from the Arkansas Razorbacks with a Framing+ of 111.47 (relative to the Ivy League average), compared to the best Ivy catcher, Robert Ciulla of Yale with a Framing+ of 108.87. If I reset Framing+ to the Division I mean, the average Ivy League catcher has a Framing+ of 98.51, Ciulla has a Framing+ of 107.38, Helfrick has a Framing+ of 109.95, and Zach Lass of the University of Richmond led all of college baseball in 2024 with a Framing+ of 112.12 (1905 pitches caught).

Zach Lass - Baseball

At Princeton, we are trying to take these Framing+ stats and delve deeper to determine which pitchers, which pitch types, and which parts of the zone our catchers excel or struggle with. Catching coach KJ Hallgren programs the pitching machine to mimic the most troublesome pitches for each catcher (a post on this topic is coming soon), and then pairs slow-motion video analysis with Trackman data to create individualized drills to improve receiving skills.

Compared to MLB, the Ivy League strike zone is a wild beast. We are just starting to have enough data to find competitive advantages hidden in this jungle.