How Umpire Scorecards Are Calculated

Everyone in the ballpark saw the same pitch. Half of them are certain it was a strike. The pitch-tracking unit behind home plate recorded exactly where the ball crossed, to within a fraction of an inch, and it has no opinion at all — which is why a scorecard is useful and also why it is so easy to misread.

An umpire scorecard turns a game of called pitches into four numbers. Most of the arguing happens over the first one, which is the least informative of the four.

This is the part most people get wrong before any math happens. A scorecard only grades pitches the hitter did not swing at — called strikes and called balls. Every swing, every foul, every ball in play is thrown out before the first percentage is computed.

That is not a shortcut. It is the only fair way to do it. The umpire never had to make a decision on a pitch the hitter swung at, so grading him on it would be grading him on someone else's choice. A 300-pitch game might produce only 160 judged calls.

The rulebook zone is the width of the plate, from the hollow beneath the kneecap to the midpoint between the shoulders and the belt. Tracking data gives two numbers per pitch: how far left or right of center it crossed, and how high off the ground.

Turning those into a verdict requires two decisions, and they are worth understanding because they change the answer.

• The plate is 17 inches wide, so its edges sit 8.5 inches either side of center. A baseball has a radius of about 1.45 inches, and a pitch counts as a strike if ANY part of it touches the zone — so the usable horizontal boundary is about 9.95 inches, or 0.83 feet, from the middle.
• The same logic applies vertically. The zone runs roughly 1.5 to 3.5 feet off the ground, and a ball clipping the top or bottom edge is still a strike, so the effective boundaries are about 1.38 and 3.62 feet.

That ball-radius allowance is not generosity toward the pitcher. It is the rule. A pitch that catches the very top of the zone with the bottom of the ball is a strike, and a system that measured only the ball's center would call it wrong and then blame the umpire for it.

The vertical zone above is a fixed band. The real one is not — it is defined by each hitter's stance, so a six-foot-five first baseman and a five-foot-eight middle infielder have genuinely different zones, and an umpire adjusting for that is doing his job correctly.

Broadcast-level systems scale the zone per batter. Most college and high school exports do not carry reliable batter height or stance data, so any scorecard built from them is using a standard band instead.

The practical consequence: trust the horizontal numbers more than the vertical ones. A pitch three inches off the outside corner is off the corner for everybody. A pitch at the letters is a judgment call that depends on whose letters.

Accuracy is the simple one. Of every pitch the hitter took, how many were called correctly? Correct calls divided by total decisions.

It is also the number most likely to flatter an umpire, because most taken pitches are not close. A fastball that bounces and a slider that misses by a foot are both decisions, and both are free. An umpire who happens to work a game with few borderline pitches will post a better accuracy figure than one who worked a game full of them, without being any better at his job.

Splitting it in two is where it gets interesting. Accuracy on pitches that were actually in the zone tells you whether he is squeezing. Accuracy on pitches that were actually out of it tells you whether he is expanding. An umpire at 95% on balls and 80% on strikes is not inconsistent — he has a small zone, consistently, and the pitching staff needs to know that by the second inning.

Consistency ignores every easy pitch and grades only the hard ones — the pitches that crossed within about one ball width, roughly 2.9 inches, of an edge. Those are the only calls where a reasonable person could disagree.

This is the fairest single measure of an umpire, and it is always lower than accuracy. A game at 92% accuracy might sit near 70% on the borderline pitches, and that is normal rather than damning. What you are looking for is whether the misses fall on both sides or pile up in one place.

Counting missed calls treats them all the same, and they are not remotely the same. A called strike on ball one is a minor annoyance. The same call on 3-2 ends the at-bat.

Favor fixes that by converting each missed call into runs, using how much the count was worth before and after. The weights rise steeply with leverage:

• A miss on 0-0 is worth about 0.06 runs.
• A miss on 1-1 is worth about 0.09.
• A miss on 3-1 is worth about 0.18.
• A miss on 3-2 is worth about 0.25 — roughly four times the cost of the same mistake on the first pitch.

Each miss is then credited to whichever side it helped, and the two totals are netted. A game can have eight missed calls and a favor of almost zero, because they cancelled out. Another can have four and a favor of a full run, because three of them landed on the same team in two-strike counts.

Favor is the number that answers the question people are really asking, which is not "was he good" but "did it cost us".

A scorecard is a measurement, not a verdict. Three things worth holding onto:

• Small samples swing hard. A seven-inning college game might contain forty borderline pitches. Four coin flips going the wrong way moves consistency by ten points.
• A tight zone called the same way all night is a good umpire, even if the accuracy figure is unremarkable. Pitchers can adjust to a small zone. They cannot adjust to a moving one.
• The vertical edges carry the uncertainty, for the reason described above. Build an argument on the corners, not the letters.

Used well, a scorecard stops the argument in the dugout rather than starting one. The catcher wants to know where the zone is tonight, not whether the umpire is any good — and the first inning of data answers that question better than three innings of guessing.