Dr. Ryan Jessup and Dr. Don Pope have teamed up to create the Jessup Pope NCAA Division I FBS rankings. Jessup, Assistant Professor of Marketing, teaches marketing research and data mining and Pope, Associate Professor of Management, teaches statistics and intro to management. These two great statistical minds decided to put their knowledge and experience to practical use in devising a new method of ranking college football teams. The blog today is written by Dr. Jessup as our COBA Blog guest writer.
I present to you the inaugural Jessup Pope NCAA Division I FBS rankings. This is a ranking of all 128 college football teams in the bowl subdivision based on the performances this season to date.
A brief history
A few years ago, after the wildly improbable success of our faculty/staff intramural soccer team, the Sunflowers of Death (which is a topic for another time, but suffice it to say that we have finished 2nd place in 4 out of the 7 seasons we have played), we wanted to come up with a better way to rank sports teams.
With the help of our then-student worker, Amy Morris, we collected the data from multiple seasons of a variety of sports. Don Pope and I then took the Google PageRank algorithm, the algorithm that jump-started the search engine giant – and still underlies their current approach today – and made several modifications to it to allow it to rank teams, instead of webpages. We also enabled it to account for home field advantage and temporal decay of performances (this is where games early in the season weigh less in the rankings than more recent games), among other things. We then forecasted bowl games at the end of three seasons, finding that our algorithm correctly predicted the winner (1) more than 60% of the time, (2) better than the BCS (or its replacement, the College Football Playoff), and (3) even out-predicted the Las Vegas betting line 56% of the time. We presented this work at the Christian Scholars Conference in Abilene this past summer. More recently, with the help of several faculty in the SITC, most prominently Dr. Ray Pettit, I wrote a screen-scraper in python (a programming language) to scrape off all the information for each college football game so that we can begin looking at our rankings on a weekly basis.
The effectiveness of the ranking system lies in the fact that it not only takes into account your team’s performance against other teams but also the other teams’ performance against other teams. So let’s take a quick look at these initial rankings. Some of these rankings accord with our expectations and others outrageously violate them. On one hand, quite reasonably, Notre Dame and Clemson are 1 and 2, and Alabama, Stanford, Michigan State, and Florida are all in the top 10. On the other hand, USC is No. 3, Texas is No. 16, whereas Ohio State and Baylor are No.’s 11 and 60, respectively, all apparently crazy! Given that they were just punished by Iowa State, I don’t believe that Texas is the 16th best team in the nation, but that aberrant win over Oklahoma at a neutral site is currently making Texas look very good. And what’s the deal with Baylor and Ohio State? Well, part of the issue is that Baylor and Ohio State are both in power conferences and began the season ranked in the top 5. Yet our system cares neither about which conference your team is in nor is it biased by preseason rankings: it only takes into account your team’s performances and the quality of the opponent (as judged by their performances against their opponents). So, apparently they have not played very strong opponents thus far.
I think as the weeks continue our rankings will begin to accord a bit more with our expectations but I won’t be terribly surprised if at the end of the season there are still some outrageous-looking rankings. You don’t beat the Vegas line 56% of the time by just heeding the status quo!
Here are our rankings at this time:
|Rank||After Week 9|
|54||San Diego State|
|110||San Jose State|
|123||New Mexico State|