ABS Umpires

The Anatomy of a Missed Call

379,155 called pitches, a 7.2% miss rate, and the half-inch cliff where human judgment breaks down

· 8 min read

Every MLB game has about 11 wrong ball/strike calls. Not borderline judgment calls that reasonable people could argue about. Wrong calls — pitches where the umpire and the rulebook zone clearly disagree.

We applied the ABS strike zone model to the full 2025 MLB Statcast dataset: 729,827 total pitches, of which 379,155 were called pitches (balls and called strikes — not swings, fouls, or hit-by-pitches). Then we checked how often the umpire's call matched the ABS zone.

The overall disagreement rate: 7.2%. That translates to 10.9 wrong calls per game on average, with a median of 11 and an interquartile range of 8 to 13.

Eleven wrong calls sounds like a lot. But the story isn't the number — it's where those misses happen and when they matter.

The Accuracy Cliff

Most missed calls aren't egregious. They're clustered in a narrow band right at the zone edge. And the accuracy degradation isn't gradual — it's a cliff.

Distance from zone edge Umpire accuracy
More than 3 inches98.6%
2 – 3 inches92.4%
1.5 – 2 inches85.3%
1 – 1.5 inches78.9%
0.5 – 1 inch68.6%
Less than 0.5 inches60.1%

When a pitch lands within half an inch of the zone edge, the umpire is barely better than a coin flip. Within that narrow band — roughly 9 pitches per game — the human eye simply cannot reliably distinguish a ball from a strike. Above 3 inches from the edge, umpires are nearly perfect. The transition between those two regimes is sharp and consistent.

The Accuracy Cliff

50%60%70%80%90%100%50% (coin flip)98.6%>3 in92.4%2-3 in85.3%1.5-2 in78.9%1-1.5 in68.6%0.5-1 in60.1%<0.5 inDistance from zone edgeAccuracy

Umpire accuracy by distance from the zone edge, in half-inch bins. The cliff begins at 1.5 inches and becomes a coin flip below 0.5 inches.

Umpires Are Generous, Not Strict

Here's something the data reveals that announcers rarely mention: umpires are systematically more lenient than the ABS zone, not more strict. Per game, there are on average 6.3 missed strikes (umpire calls ball, ABS says strike) versus 4.6 false strikes (umpire calls strike, ABS says ball).

The human zone is slightly smaller than the rulebook zone. Umpires give batters the benefit of the doubt on close pitches, particularly at the top and bottom edges. This is a consistent, league-wide pattern — not an artifact of a few lenient umpires pulling the average.

Not All Missed Calls Are Equal

A wrong call on a 1-0 count changes the count to either 2-0 or 1-1 — a meaningful but modest shift. A wrong call on a 3-2 count is either a walk or a strikeout — a swing of 0.690 wOBA, the single most consequential ball/strike decision in any at-bat.

We computed the counterfactual wOBA impact of a wrong call at every count — the difference in expected offensive value between the two possible outcomes:

Count Challenge value (wOBA)
3-20.690
2-20.384
3-10.306
1-20.273
0-20.230
2-10.210
3-00.207
2-00.195
1-10.131
1-00.130

A wrong call at 3-2 is more than 5 times more valuable than one at 1-0. This has direct implications for the ABS challenge system: if a team has one challenge left, the count should heavily influence the decision to use it.

Challenge Value by Count

3-2
0.690
2-2
0.384
3-1
0.306
1-2
0.273
0-2
0.230
2-1
0.210
3-0
0.207
2-0
0.195
1-1
0.131
1-0
0.130
0-1
0.118
0-0
0.095

wOBA impact of a wrong call at each count — full counts are 5× more valuable than early counts

Counterfactual wOBA value of a wrong call by count. Full counts are 5x more valuable than early counts.

Where Umpires Miss: The Zone Heatmap

Loading...

Distribution of missed calls across the strike zone. Red = false strikes (called strike, actually ball). Blue = missed strikes (called ball, actually strike). Green = correct calls.

The Catcher's Hidden Skill

One more finding, and it's the one that surprised us most. On borderline pitches — those within 1.5 inches of the zone edge — the identity of the catcher dramatically influences whether the umpire calls a strike.

The best framers in the game earn a called strike on 51.7% of borderline pitches. The worst: 33.0%. That's an 18.7 percentage point gap on the most consequential pitches of every game.

In the ABS era, this creates a fascinating strategic tension. Good framing may lose value because bad calls get overturned on challenge. But good framing may gain value because it makes the umpire's initial call look correct to the batter deciding whether to challenge — if the pitch looks like a strike coming in, the batter is less likely to tap his helmet even if it was technically outside.

What This Means for the ABS Era

Starting on Opening Day 2026, MLB introduced the challenge system powered by ABS. Every game now has a mechanism to correct the most consequential missed calls. But the data suggests the story is more nuanced than "robots fix everything":

  • Most missed calls are borderline. The 60.1% accuracy zone is a half-inch band. These aren't egregious misses — they're pitches that genuinely look like they could go either way.
  • Challenge strategy is count-dependent. Teams should strongly prefer to challenge on two-strike and full counts, where the wOBA swing is largest.
  • Catcher framing still matters. Even with challenges available, the umpire's initial call shapes whether the batter challenges at all.

CalledThird will track umpire accuracy, challenge decisions, and framing impact throughout the 2026 season. Every game, every call, every challenge — automated and data-driven.

Methodology

Data source: 2025 MLB Statcast data (729,827 pitches, 379,155 called pitches). Accessed via pybaseball.

Zone model: ABS zone specification follows the official 2026 rules: 17-inch plate width, batter-specific height boundaries (sz_top, sz_bot from Statcast), strike credited if any part of the ball (1.45-inch radius) touches the zone.

Distance calculation: Minimum distance from pitch center to nearest zone edge, computed in 2D (horizontal and vertical). Binned in half-inch increments.

Challenge value: Counterfactual wOBA computed as the difference between the expected wOBA of the two possible count transitions. For terminal counts (3-2), this is the difference between walk and strikeout outcomes.

Framing analysis: Called strike rate on borderline pitches (within 1.5 inches of zone edge) grouped by catcher, minimum 200 borderline pitches received.

If you find an error, tell us — we'd rather be corrected than wrong.

Full methodology documentation →