Tuesday, November 27, 2007

Sports Myths

Reason #839 why I love economics: So many economists publish papers dealing purely with sports data. I've illustrated this once to help diffuse the myth of the value of punting.

The latest "sports economist" to gain attention is Trevor D. Logan at The Ohio State University. His new paper's title is: "Whoah, Nellie! Empirical Tests of College Football's Conventional Wisdom." (Hat tip: Freakonomics).

Logan uses AP poll data along with 25 years of data from 25 different teams to examine the following widely-quoted conceptions about polls and declares them myths:

1. It's better to lose early in the season than late.
2. Voters reward wins over tough opponents.
3. Blowing out your opponent is rewarded more than narrowly beating them.

Logan finds that:
1. Voters re-rank teams that lose later higher than they would have if they had lost earlier in the season.
2. Strength of your opponent only matters if you lose. Losing to a tougher opponent allows you to be ranked (slightly) higher than you would be had you lost to a bad opponent.
However, beating a good opponent doesn't affect your ranking more than beating a bad opponent.
3. Likewise, blowouts only matter if you lose. Blown-out teams are ranked lower than narrow-loss teams. But, the winning teams in blowouts are not ranked higher (or lower) or rewarded for throttling the other team.

I'll compare Logan's working paper with Gene Wojciechowski's ESPN column today about "BCS myths." Wojo uses no data, only his opinion, but some of them are funny:

Wojo's myths are:
1. The BCS Works.
We're stuck with the BCS and its weekly standings weirdness. For example, Missouri is your No. 1 team in the country. This is like Homer Simpson picking up Eva Mendes at a Chi Omega party. Nothing against Mizzou and quarterback Chase Daniel, but the Tigers aren't the No. 1 team in the country. They aren't even favored in Vegas to beat No. 9 Oklahoma in the Big 12 championship (Mizzou is a 3-point dog).


2. Heisman Trophy voters know what they're doing.
There are 925 Heisman voters -- 870 media, 54 living Heisman winners, one collective fan vote. The more voters, the more probability of the dreaded Knucklehead Factor... Twenty years ago, a Heisman voter once told me he never cast a first-place vote for an African-American player
3. Nebraska is an elite coaching job.


I find the regression analysis to be more enlightening than just a pundit's opinion. One asks "What does the data say?" while the other just says what he thinks.

No comments: