December 19, 2008

Conference Debates: BCS Criteria

With the publication of the BCS's criteria for determining future automatic qualification it is amazing that, until now, no one has published numbers to see what the BCS looks at.

The methodology used to generate these numbers can be found here, where it can also be found that they are mine ; ) Basically 0.5000 is designed to indicate the line between BCS merit and those who come up short.

For 2008 the approximated BCS conference rankings are:

Big 12: 0.9819
SEC: 0.7892
ACC: 0.7455
Big 10: 0.6689
MWC 0.5347
Big East: 0.5259
PAC 10: 0.4069
WAC: 0.2993
MAC: 0.1144
C-USA: 0.0020

These rankings, like popular opinion, are biased towards the top of the standings. This is different from conference rankings like Sagarin's, or even the computer component of this data, that give a better measure of the middle of the conference.

The BCS uses a 4 year window. Here are the rankings over this window:

SEC: 0.8974
Big 12: 0.8087
Big 10: 0.7989
ACC: 0.7265
PAC 10: 0.6686
Big East: 0.6058
MWC: 0.2627
WAC: 0.2386
Independents: 0.1468
MAC: 0.0287
C-USA: 0.0030
Sun Belt: 0.0000

Upon examining the data it turned out that I was wrong to shoot down the idea at the start of the season that the WAC in recent years had out performed the MWC, though this year made up for that error on my part.

This would indicate that the MWC's performance this year would merit BCS inclusion if it could be maintained over a 4 year period. With the addition of Boise State and removal of the 2005 data the MWC would move up to 0.4250, on the cusp of warranting a BCS automatic qualification. They would need a year better than this year to merit consideration next year, or slightly less than this year over the next two years.

The numbers also show that the PAC 10 with the addition of BYU and Utah would have a lower average over this window, with 2008 being an exception.

Comments:

Conference Debates: BCS Criteria
With the publication of the BCS's criteria for determining future automatic qualification it is amazing that, until now, no one has published numbers to see what the BCS looks at. The methodology used to generate these numbers can be found here, where it can also be found that they are mine ; ) Basically 0.5000 is designed to indicate the line between BCS merit and those who come up short. For 2008 the approximated BCS conference rankings are: Big 12: 0.9819 SEC: 0.7892 ACC: 0.7455 Big 10: 0.6689 MWC 0.5347 Big East: 0.5259 PAC 10: 0.4069 WAC: 0.2993 MAC: 0.1144 C-USA: 0.0020 These rankings, like popular opinion, are biased towards the top of the standings. This is different from conference rankings like Sagarin's, or even the computer component of this data, that give a better measure of the middle of the conference. The BCS uses a 4 year window. Here are the rankings over this window: SEC: 0.8974 Big 12: 0.8087 Big 10: 0.7989 ACC: 0.7265 PAC 10: 0.6686 Big East: 0.6058 MWC: 0.2627 WAC: 0.2386 Independents: 0.1468 MAC: 0.0287 C-USA: 0.0030 Sun Belt: 0.0000 Upon examining the data it turned out that I was wrong to shoot down the idea at the start of the season that the WAC in recent years had out performed the MWC, though this year made up for that error on my part. This would indicate that the MWC's performance this year would merit BCS inclusion if it could be maintained over a 4 year period. With the addition of Boise State and removal of the 2005 data the MWC would move up to 0.4250, on the cusp of warranting a BCS automatic qualification. They would need a year better than this year to merit consideration next year, or slightly less than this year over the next two years. The numbers also show that the PAC 10 with the addition of BYU and Utah would have a lower average over this window, with 2008 being an exception.