Baseball Musings
Baseball Musings
October 08, 2007
Diamondbacks in the Rough

AZ Snakepit argues that the Diamondbacks are good, not lucky.

If the Diamondbacks were a 79-win team, as Pythagoras believes, this binomial calculator gives them only a 4.9% chance of reaching 90 wins. Item #2. Baseball Prospectus predicted a 7.0% chance of us sweeping the Cubs. Combining these two, the odds of both happening is one in 289. Really, there's a point at which blaming "luck" in the face of relentless evidence goes beyond stubborness, and begins to drift into idiocy. A good scientist, when the evidence piles up against them, will admit that their theory is flawed, and will begin to look around for a better theory. Jay, however, seems to feel that when the facts disagree with his theory, the facts must be disposed of. Blaming "luck" for an 11-game differential - especially two seasons after the same manager, for the same franchise, posted exactly the same improvement - is like continuing to play poker after your opponent gets three royal flushes in a row. There's a certain point beyond which it's time to check the deck.

I've checked the deck, and there's nothing there that indicates this team is good. That doesn't mean they can't win. And it doesn't mean they can't continue to be lucky.


Posted by David Pinto at 05:02 PM | Team Evaluation | TrackBack (0)
Comments

1 in 289 are actually pretty good odds. That's not exactly rare.
Other things that are 1 in 289:
Homes forecloses in Arizona for 2007
Number of people over 5 yrs. who speak Cajun at home in Louisiana
Having a kid with Down Syndrome when mom is 36

I guess foreclosure, Cajun and Downs are all too incredible to believe.

http://www.google.com/search?hl=en&safe=off&q=%221+in+289%22&btnG=Search

Posted by: Harry at October 8, 2007 06:30 PM

Yeah, 4.9% hardly means it's impossible. But while a lot of these tools are good, when a team performs better or worse than expected, there can be a lot more explanations than just luck. A Pythagorean W/L isn't a law of physics or anything. It could easily be skewed by having a handful of really poor pitchers, for example.

Viz., a simplified model of a team that reliably scores exactly 4 runs each game, has 3/5 of its pitchers allow exactly 3 runs each game, and 2/5 of them allow exactly 6 runs each game. Such a team would, in fact, win 60% of its games, but would have a 77-85 Pythagorean record. Obviously no team is always going to score the same number of runs or anythnig like that, but the point is that a Pythagorean W-L assumes that a team's allowed and scored runs both follow a normal distribution. Deviations from such a distribution may be luck, or they may be real.

Posted by: Adam Villani at October 8, 2007 09:35 PM
Post a comment









Remember personal info?