The BCS can't seem to get the monkeys off of it's back. In particular, there was recently an opinion piece in the Washington Post that discusses some work in which I've been involved (with Thomas Callaghan and Peter Mucha; Thomas was the one who did the real work; Peter and I were advisors on tje project) about using random walkers (on the graph defined by the college football schedule) to rank football teams. The short version is that our random walkers, which we playfully dubbed "monkeys", do just as well as the BCS. The Washington Post article proposes the idea of replaces the experts with monkeys, which is actually kind of reminiscent about how things were jokingly phrased when we first came up with the monkey analogy (which is actually my fault, because I would come up with comments asking whether monkeys can do as well as the experts).
I mentioned this project in an earlier blog entry, but I'm reiterating it because of the Washington Post gig. Our 2005 rankings our available on the official project website, from which our papers can also be downloaded. We've now taken to using the less snide and more accurate term 'random walkers' in our presentations, although the monkeys still get mentioned now and again for fun, especially as we can't exactly escape them at this point. (The official webpage was originally called "Monkeys Shock the BCS" and you know I am the one who wanted to allude to Peter Gabriel.)
There were rumors floating around (which might have just been somebody being nice to us; I only heard this third-hand) that our system was actually considered by the Powers That Be to be added to the official BCS system but that the use of the term 'random monkeys' hurt our chances---hence the change to random walkers. Of course, with the article in the Washington Post, all such chances are probably dead and buried. But maybe we'll get used in the Congressional hearings? That would be pretty cool...
3 days ago
2 comments:
This is great stuff Mason. I'm impressed.
And I should add I'm a fan of any system that ranks UT over USC.
What the media and nearly everyone else has missed this year is how good our defense is. Unless the Trojans switch to the option- a la Texas A&M- I don't see them hanging 50 or more on us. USC didn't play anybody who had a defense as good as ours all year.
And I think we can hang at least 50 on them, since their defense is highly suspect.
Guess we'll find out 4 Jan.
Hook 'em Horns.
Then you can protect me from my USC-homer cousin, who actually transferred there from UCLA. He broke several hearts---the other four people in my family are all UCLA alums. I'm the only deviant (again).
Of course, our system knows nothing about the defenses of the teams. There are a few parameter value for which USC comes out first (p near 1), but the random walkers pick U Texas most of the time, and it is likely for strength-of-schedule reasons (which becomes less important for the method as p becomes larger).
Post a Comment