Wednesday, April 06, 2005

Bill James and explaining vs explaining away

Via Dan Drezner and Mark Kleiman, Bill James takes a less-than-absolutist tack on sabermetrics. Kleiman notes that, since New York and Boston were evenly matched, the Yankees' 4-in-a-row collapse against the Red Sox was 1/16 likely from statistics - he thinks that James's "veteran leadership" claim for why this happened is an example of the fundamental human tendency to latch on to explanations to random events. However, my guess is that James's point is that the RS didn't suffer a failure of nerve, which would have lowered their chances significantly below statistics - presumably "evenly matched" includes a tacit "in games played in balanced emotional states". In fact, I conjecture that an analysis of playoff series made under the "evenly matched' assumption will find a correlation between the late results and the early, and that even adjusting for expected team strength, winning the first few games gives a team a psychological edge, esp. against some class of teams. That would be "teams without veteran leadership" or "teams not made up of youngsters too naive to be rattled by being down" or - well, now I'm in bump-hunting territory.



Post a Comment

<< Home