The MVP Machine: How Baseball's New Nonconformists Are Using Data to Build Better Players

by Ben Lindbergh

Before Bannister, the closest the statistical community had come to influencing a major-league pitcher’s performance was what came to be called the “Félix Incident.” In June 2007, future FanGraphs managing editor and Padres analyst Dave Cameron—then a full-time corporate cost analyst for Hanes and a part-time blogger for the Mariners site U.S.S. Mariner—posted “An Open Letter to Rafael Chaves,” the Mariners’ pitching coach. Cameron observed that the Mariners’ ultratalented but mercurial ace-in-waiting, Félix Hernández, was far too fastball-reliant early in games, and he pleaded with Chaves to make him mix it up. Soon after, a fan in the stands handed a printout of the post to Chaves. Chaves, who claimed that he’d already been trying to persuade Hernández to vary his pitch selection, relayed the post to the pitcher to prove that his pattern was well known. “Chaves gave me a report,” Hernández said in early July, after his second consecutive strong start. “On the Internet, they say when I throw a lot of fastballs in the first inning, they score a lot of runs. I tried to mix all my pitches in the first inning.”