NCAA Division 1 Football Rankings

After toying around with the idea for a few years I've finally done and created a College Division 1A ranking model.  It is a bit more complicated and time consuming than the NFL version.  This is why it took five years to get around to it.

While I like this first version I think I may still tweak the subjective weights.  But all in all I like the results.  Part of what I like is that there is both overlap and contradictions with the accepted "wisdom" of the national voters and the junk BCS ratings.

On to my top 25:

RANK	SCHOOL	RECORD		CONFERENCE	OFFENSE RANK	DEFENSE RANK
1	FLORIDA ST	13 - 0		ACC	2	1
2	BAYLOR	11 - 1		BIG 12	1	7
3	ALABAMA	11 - 1		SEC	11	3
4	OHIO ST	12 - 1		BIG 10	3	32
5	LOUISVILLE	11 - 1		AMERICAN	14	4
6	AUBURN	12 - 1		SEC	5	51
7	MICHIGAN ST	12 - 1		BIG 10	32	2
8	ARIZONA ST	10 - 3		PAC 12	6	11
9	OKLAHOMA ST	10 - 2		BIG 12	13	6
10	MISSOURI	11 - 2		SEC	9	14
11	CLEMSON	10 - 2		ACC	18	12
12	OREGON	10 - 2		PAC 12	7	22
13	UCF	11 - 1		AMERICAN	27	33
14	NORTHERN ILLINOIS	12 - 1		MAC	8	82
15	S CAROLINA	10 - 2		SEC	17	34
16	FRESNO ST	11 - 1		MOUNTAIN WEST	23	66
17	BOWLING GREEN	10 - 3		MAC	25	20
18	STANFORD	11 - 2		PAC 12	33	31
19	WISCONSIN	9 - 3		BIG 10	19	19
20	BALL ST	10 - 2		MAC	26	64
21	OKLAHOMA	10 - 2		BIG 12	41	16
22	E CAROLINA	9 - 3		CUSA	15	47
23	MIAMI	9 - 3		ACC	28	49
24	DUKE	10 - 3		ACC	50	40
25	RICE	10 - 3		CUSA	34	39

Now if it were up to me, Florida State would be playing Baylor in the final game.  Not only is Baylor the only offense that has statistically outperformed FSU, but Baylor lead the NCAAs until their loss a couple weeks back. 

Also for my Big 10 folks, note that MSU and Ohio State are mirror images of each other statistically.  It would be great to see them play again.

Now the first reason why this is more complicated than the NFL ranks on my blog is that there are so many teams.  Unlike the 32 in the NFL there are 126 Division 1 schools in the NCAA.  In order for my model to work I needed all data from all schools in all games. 

The second issue is that nearly all schools play at least one game against a division 1-AA school at some point.  Now in order to properly generate a weight for a 1 AA school I would in theory need to keep track of all 1 AA schools so that they would have proper data.  Now you see how this gets more cumbersome.  The work around I employed was to average all statistical measures of 1 AA schools to generate the single game statistics weights.  Generally these schools did rather poorly.  But strange enough, statistically the 1AA schools were as a group not the worst in the collegiate ranks.

Incidentially the BCS top 25 are this:

1	Florida State
2	Auburn
3	Alabama
4	Michigan State
5	Stanford
6	Baylor
7	Ohio State
8	Missouri
9	South Carolina
10	Oregon
11	Oklahoma
12	Clemson
13	Oklahoma State
14	Arizona State
15	UCF
16	LSU
17	UCLA
18	Louisville
19	Wisconsin
20	Fresno State
21	Texas A&M
22	Georgia
23	Northern Illinois
24	Duke
25	USC

Texas A&M, LSU, USC and UCLA are all noteably overrated by the BCS according to my system, just as Louisville, Baylor, Arizona State are all underrated by the BCS.  After looking at every box score from 126 teams (most of which having played twelve games), I like my results better.

Part of my goal here was to be able to more fully appreciate which conferences rank the best.  And in this case the public opinion does match the results.

The SEC has, in my system, the highest average ranking position at 48.00. 

	CONFERENCE	AVE SCHOOL RANK
1	SEC	48.00
2	BIG 10	51.75
2	PAC 12	51.75
4	ACC	56.93
5	BIG 12	59.60
6	AMERICAN	70.20
7	CUSA	72.50
8	MOUNTAIN WEST	72.75
9	MAC	73.46

It was an interesting exercise.  The results were both surprising and not at the same time.  And that is all I could hope for.