Brian O'Nora

What we can't say about Brian O'Nora

This season is 2,404 graded takes, which puts Brian O'Nora's overall accuracy at 92.5% ± 1.0 (95% interval). That interval is a large share of the whole league's four-point spread, so this grade is a reading and not yet a ranking — we print it, and we do not place it against umpires with a full season behind the plate. Everything finer is further out of reach still:

• how the zone moves with two strikes — about 481 takes, so ±2.3 points. (the league-wide two-strike effect is about 1 point in 2026.)
• any single corner of the zone — about 200 takes, so ±3.6 points. (corner cells are where umpires actually differ.)
• whether one pitcher is called tighter than another — about 40 takes, so ±8.1 points. (a given pitcher is worked once or twice a season.)

And a note on the number itself: an accuracy figure is a rate plus a convention for what counts as a miss. Ours grades every taken pitch against the rulebook zone as the feed reports it. Others are more forgiving — UmpScorecards counts a call wrong only when the pitch was at least 90% likely to be a strike — so two sites can print the same percentage about the same umpire and not mean the same thing by it. How the grade is built.

The overturn rate is a different kind of number and does not belong beside the grade: it is accuracy on the 70 pitches players chose to challenge, which are the pitches they already believed were wrong. 70 is enough to print it with its own denominator, which is what we do — but it still ranks umpires by how often they are appealed, not by how often they are right.