An umpire calls eight strikes on ten borderline pitches when a model expected six, which proves nothing about a wide zone. Any umpire strike zone claim needs a better yardstick than raw averages, because umpires draw different games and different pitchers. You will learn to count taken pitches, estimate a called-strike probability from location, count and batter, then add up observed minus expected as a residual. Catcher framing gets the same treatment, with a warning that a catcher receiving tough pitches can look worse than he is. Ten pitches show nothing lasting, so the lesson stresses sample size, intervals and out-of-sample checks. Use it to tell a real edge in totals or strikeout props from a story.
Learning objectives
Interpret the technique, reproduce the example, describe limitations, and apply it to a simulated research workflow.
