Each week during the season I plan to rank the top 25 NCAA FBS teams using my
Complex Invasion College Football Production Model. Now that we have two weeks of data, the model does a fairly good job of estimating "offensive production" and "defensive production", so I have decided to post the first of the top 25 ranks for this past week. I had a lot of changes to make (teams switching conferences - and how that was coded in the data along with the addition of Georgia State). I am still working on the strength of schedule stuff as the source that I used in the past is no longer reporting the data, so I will either have to find another source or start entering it myself.
In the future I plan on having the top 25 done on Monday morning, with a small delay for this week. So, without any more delay - here is the Top 25 for this past weekend.
FYI:
2013 NCAA FBS Top 25 for Week 3
| Rank |
Team |
| 1 |
Baylor |
| 2 |
Oregon |
| 3 |
Maryland |
| 4 |
Wisconsin |
| 5 |
Houston |
| 6 |
Georgia Tech |
| 7 |
Louisville |
| 8 |
Marshall |
| 9 |
UCF |
| 10 |
Duke |
| 11 |
Tennessee |
| 12 |
Florida State |
| 13 |
Arizona State |
| 14 |
Utah |
| 15 |
Michigan |
| 16 |
Wyoming |
| 17 |
LSU |
| 18 |
Ohio State |
| 19 |
Texas Tech |
| 20 |
Boston College |
| 21 |
Texas A&M |
| 22 |
Colorado |
| 23 |
Michigan State |
| 24 |
Texas State |
| 25 |
Arizona |
Alabama is pretty good too. You don't have them.
ReplyDeleteYes, Alabama should be very productive, but is not in the top 25. Why? The model looks at how productive teams are against their opponents and how that ranks relative to how other teams perform against their opponents. Since Baylor has dominated two teams, that is why Baylor is currently #1. I expect them to drop a bit as the season progresses and Alabama to rise over the next few weeks.
ReplyDeleteThis is very interesting. I look forward to watching the expected volatile adjustments over the next few weeks as we get closer to a statistically significant sample size. Will you be releasing your full list including the underlying values behind the rank? Among many interesting features that would reveal would be if there are any major gaps between teams/cohorts.
ReplyDeleteSteve, as of now I plan on releasing each week's top 25 as I have done over the last few years. Here is last year's final (along with links to the previous weeks) http://teamsportsanalysis.blogspot.com/2013/01/2012-final-top-25.html.
ReplyDeleteI can release the full list as well, and the regression results. I will try to get this done in the future.