http://fivethirtyeight.com/features/fou ... youre-out/
Consider a forgotten game in April 2010 between the Cleveland Indians and the Chicago White Sox. The White Sox were up a run with two outs in the eighth. Their set-up man, Matt Thornton, was on the mound, protecting a lead with a runner on first and the right-handed Jhonny Peralta at bat. Ahead in the count with one ball and two strikes, Thornton froze Peralta with a slider on the outside half of the plate, a couple inches below the belt. For a pitch like that, the umpire, Bruce Dreckman, would normally call a strike — 80 percent of the time, the data shows. But in two-strike counts like Peralta’s, he calls a strike less than half the time.
Sure enough, that night Dreckman called a ball. Two pitches later, Peralta lashed a double to right, scoring the runner and tying the game. Neither team scored again until the 11th, when Cleveland scored twice to win the game. Had Peralta struck out to end the top of the eighth, Chicago almost certainly would have won.1
This one call illustrates a statistical regularity: Umpires are biased. About once a game, an at-bat ends in something other than a strikeout even when a third strike should have been called. Umpires want to make the right call, but they also don’t want to make the wrong call at the wrong time. Ironically, this prompts them to make bad calls more often.
That’s according to research I did with David P. Daniels showing that the strike zone changes when the stakes are highest. We looked at more than 1 million pitches, almost all ball and strike calls from the 2009, 2010 and 2011 regular seasons, and found that the strike zone expands in three-ball counts and shrinks in two-strike counts.2 It also shrinks again when the preceding pitch in the at-bat was a called strike. To put it another way, on close calls, umpires are unlikely to call a fourth ball, a third strike, or a second strike in a row. Umpires call balls and strikes as if they don’t want to be noticed.