Thursday, September 29, 2011

Now I can pay attention to football...

Though my passport from the Red Sox Nation has faded in a decade away from Boston living, I still feel gutted by the just completed September swoon.  Nate Silver beautifully quantifies the epic level of the Red Sox collapse this year.  Even as he notes that the estimate is not at all mathematically rigorous, it's a twisted kind of fun to have a numerical narrative building to the claim of "about one chance in 278 million of all these events coming together in quite this way."  For even casual Red Sox fans, those numbers are believable in our gut.

Looking for the silver lining on this cloud, I now plan to see whether I can get the Random Walker Rankings for the 2011 college football season running soon...



At September 29, 2011 at 9:47 PM , Anonymous John Quintanilla said...

I for one am looking forward to following your rankings again this season. Thanks for all of your work on this!

At October 2, 2011 at 5:59 AM , Anonymous Ed Johnson said...

I always look forward to them too.
I tried using your system after week 3, but couldn't work out a way to get Excel to run infinite iterations. I settled on the rather arbitrary figure of 139. I also counted the whole BCS as one team, which is guess is cheating a little.
The Big 12 did very well. After week 4 they'd only notched up two non-conference losses. They were against Arizona State and Georgia Tech, who've only lost one game between them, so I'd expect to see plenty of Big 12 teams in the Top 25.

At October 2, 2011 at 11:23 AM , Anonymous Ed Johnson said...

Oops. I counted the whole FCS as one team, not the whole BCS.


At October 2, 2011 at 5:11 PM , Blogger Peter J. Mucha said...

Ed & John, thank you for the nice comments.

Ed, sorry, I don't have any tips about solving linear systems in Excel. But I will note that back when we started this project we also used the simplifying assumption of treating all of the non-FBS (then non-Div-IA) teams as a single "team" that played many games.

Eventually we moved to running on the largest connected component of teams; but this has some odd properties for our First-Last scheme, in that the vast majority of the weight of "last place" votes are outside the FBS, so we are only slightly correcting from counting only first place votes.


Post a Comment

Subscribe to Post Comments [Atom]

<< Home