Baseball Musings
Baseball Musings
November 05, 2008
Sabermetrics Wins Again

Congratulations to Barack Obama on his victory over John McCain. Congratulations, too, to Nate Silver, who called the election extremely well. It's another victory for sabermetrics, or at least the political equivalent.

After the 1990, STATS, Inc. published their annual Bill James Baseball Handbook. For the first time, that book contained Bill's projections for batters for the 1991 season. When Peter Gammons reviewed the book, one of the things he gleaned from the projections was that Bill James predicted that Jeff Bagwell was going to win the NL batting title.

James did not make that specific prediction in the book. Bagwell was traded to the Astros in September for Larry Anderson and it wasn't clear that Jeff was going to play, given the Astros had Ken Caminiti at third base. His BA projection, however, was the highest of any National League player, so Gammons was in a sense right.

Bill later told me that he felt his whole system was on the line that year. If Bagwell failed, no one would trust it again. The Astros move Bagwell to first base, however, and while he didn't win the batting title, he did hit .294 and win Rookie of the Year. James was vindicated, and the system survived.

Nate, I'm sure, faced the same kind of scrutiny with 538. He nailed the result. It's another victory for statistical analysis, and I hope this spreads to more areas of political decision making as well.


Posted by David Pinto at 07:52 AM | Statistics | TrackBack (0)
Comments

Ditto -- kudos to fivethirtyeight.com and the staff & crew. Their final pre-election projections had the electoral count at Obama 348.6 and McCain at 189.4, and the popular vote count at Obama with 52.3% and McCain at 46.2%. Currently the electoral count is 349-147 and the popular vote count is 53%-47%. I would say they nailed it.

Posted by: Joe Galloway at November 5, 2008 11:44 AM

I read www.fivethirtyeight.com before he outed himself as Nate Silver, and my mind was blown when I found out it was him. Hands down blog of the year.

Posted by: gurk at November 5, 2008 02:49 PM

Truth be told, Nate was doing things a little different from Bill James, in that he kept adjusting and adjusting his underlying logarithm.

Add to that the fact that the Dem primary season was VERY long, thus giving Nate a chance to gauge the different (regular), and frankly, Nate was not exactly at risk of his model falling apart on him (at least in the way the Bagwell / Bill James anecdote suggests).

Posted by: JRVJ at November 5, 2008 04:13 PM
Post a comment









Remember personal info?