Pre-Season Rankings

Hey football fans! Who’s ready for kickoff? I know you’re all on the edge of your seat for tonight’s Utah vs. Montana State tussle so I’d better get the pre-season rankings out quick!

It’s been an interesting offseason. Here’s a quick reminder of what we learned:

• We learned that guys that wear sweater vests can still commit major NCAA violations
• We learned that The U is still The U and that Nevin Shapiro is the new Uncle Luke
• We learned that Jordan Jefferson has 49 pairs of athletic shoes
• And just yesterday, we learned that Texas A&M would rather get beat up on by the SEC than by their older brother

 Interesting indeed! Of course, the big news that actually involves football was the offseason conference re-shuffling. The big moves were made by Nebraska, who left the Big 12 to join the Big 10, Colorado who left the Big 12 to join the Pac 12 and Utah, who took a step up from the Mountain West to the Pac 12 but BYU, who left the Mountain West and are now independent, and Boise State, who left the WAC for the Mountain West, made moves of their own that will also impact the college football landscape. These moves could have a major impact on the national championship picture but we’ll leave that discussion for later.

 I take a fully Bayesian approach to ranking college football teams, which means that I must specify a prior distribution in order to fit my model. The prior distribution represents our prior belief about the relative strength of the teams before the season starts. Many ranking systems (Jeff Sagarin’s, for example) use last season’s final rankings as a prior distribution but I don’t think this makes sense given the large amount of turnover in college football. Instead, I use a combination of published pre-season rankings that I translate into the number of points above and below average for all 120 teams. A blog post providing a full description of the prior is forthcoming. My pre-season rankings are my rankings based only on the prior. My pre-season top 10 can be found below (full rankings can be found on my personal website: www.biostat.umn.edu/~josephk/ps_rankings_2011.pdf).

Rank	Team	Rating	Record	SOS1	SOS2	E(FR)	P(WARG)
1	Alabama	24.72	0-0	3.09	0.093	9.95-2.05	0.06
2	Oklahoma	24.58	0-0	4.83	0.082	9.97-2.03	0.075
3	Oregon	23.88	0-0	3.46	0.124	10.14-1.86	0.074
4	Florida State	23.32	0-0	2.29	0.167	10.21-1.79	0.076
5	LSU	23.04	0-0	7.67	0.037	8.98-3.02	0.013
6	Stanford	22.06	0-0	2.33	0.200	10.22-1.78	0.087
7	Boise State	21.5	0-0	-4.2	0.390	10.88-1.12	0.25
8	Oklahoma State	20.8	0-0	3.78	0.105	9.5-2.5	0.035
9	Texas A&M	20.66	0-0	4.88	0.097	9.43-2.57	0.031
10	Arkansas	20.24	0-0	4.74	0.053	8.95-3.05	0.008

 As most of you know, Alabama, Oklahoma and Oregon are the pre-season favorites and all start the season about 25 points above average (note: a team’s rating can be interpreted as the number of points a team is above or below average), which is about average for the best team in the country at the end of the season. Also included in the table are two measures of strength of schedule (SOS¹ and SOS²), the expected final record (E(FR)) and the probability that they win all remaining games (P(WARG)). SOS¹ is the average rating of a team’s opponents. Most teams in the top 10 have an SOS¹ between 2 and 7.5 with the exception of Boise State, who’s SOS¹ is -4.2. To put that in perspective, the average team on Boise State’s schedule is about as good as 72 ranked Nevada. SOS² is the probability that a team that is 25 points above average would go undefeated with the given team’s schedule. This is perhaps the more relevant measure of strength of schedule for college football as going undefeated is the only sure way to assure that you will play for a national title (assuming that you’re from a BCS conference, of course). The differences SOS² among teams in the top 10 are not trivial. For example, an elite team would have only a 5% chance of going undefeated with LSU or Arkansas’s schedules, a 20% chance of going undefeated with Stanford’s schedule and a whopping 40% chance of going undefeated with Boise State’s schedule! This goes to show that even though Boise State is a very good football team they have a much easier road to going undefeated that the rest of the top 10.

Another way to look at the pre-season rankings is to consider the top 15 teams with the best chance of going undefeated:

P(WARG) Rank	Rank	Team	P(WARG)
1	7	Boise State	0.25
2	17	TCU	0.117
3	11	Virginia Tech	0.112
4	52	Houston	0.088
5	6	Stanford	0.087
6	4	Florida State	0.076
7	2	Oklahoma	0.075
8	3	Oregon	0.074
9	1	Alabama	0.06
10	24	West Virginia	0.036
11	8	Oklahoma State	0.035
12	12	Wisconsin	0.032
13	9	Texas A&M	0.031
14	13	Notre Dame	0.029
15	29	BYU	0.018

 This list looks quite a bit different than the top 10. Yes, most of the top 10 are present but teams like TCU, Virginia Tech and Houston also have a great chance of running the table even though they are not at the top of the rankings. It’s also worth noting that these calculations do include conference championship games. Conference championship games pose an extra challenge when trying to run the table because it tacks an extra game on the schedule against a quality opponent. This means that teams in the Big 10 and Pac 12 are now less likely to go undefeated than they were in the past because their conferences will play a conference championship games at the end of the season. In contrast, Oklahoma might be the biggest beneficiary of the conference reshuffling as the Big 12 will not play a conference championship game for the first time since 1995.   

 I don’t know about you but this has me excited. I hope you are too! I can’t wait until kickoff.