The problem with ranking conference strength before the bowl season is two-fold. The first obstacle is the general lack of out of conference games that pit BCS conference teams agasint each other in the early weeks of the season. The second is that once conference play begins the paradigms of college football take over, and those take a very long time to catch up to a conference's actual strength.
Thus we have to look to the bowl season to give us the true story on the relative stengths of the conferences. Of course the bowl season presents its own set of problems: dissapointed teams going to smaller bowls have a long history of being upset by a hungrier underdog. Not all BCS conferences play each other in the Bowls, and if they do it might be a game pitting six and seven seeds. Also conferences don't send the same amount of teams to bowls; last year the Pac-10 sent five teams to bowls while the ACC sent 10 programs, more teams that are even in the Big East.
Trying to put together a mathematic formula to this mess gets complicated very, very quickly. But this is the system that I have so far, and it will be based on a scale of 0-100.
Each bowl bound team is ranked according to their conference schedule only. I have disregarded the non conference games in order to rank the teams against the same level of opponents, even though the Big Ten, SEC, Big 12 and ACC do not play the same conference schedule and may played a weaker seet of teams from the other sub-conference. Furthermore each team is ranked according to the pre-bowl ranking, not their spot in the final post-season poll.
For example the Big XII from the 2008 Bowl season would rank as such:
1. Oklahoma (7-1)
2. Texas (7-1)
3. Texas Tech (7-1)
4. Oklahoma State (5-3)
5. Missourri (5-3)
6. Nebraska (5-3)
7. Kansas (4-4)
Since I don't think a linear system of ranking teams in a conference (simply saying that OU is #1 and Kansas is #7) would tell the entire story, I have a devised a system of awarding points to each team within the conference, so we can measure whether a conference is more "top-heavy" or "evenly matched". Clearly the Big XII is top heavy, with three teams at 7-1 and the other four bowl teams having close to .500 records.
This sysetm averages two numbers, the first is simply the number of conference wins. The second takes into account the teams conference ranking and its conference record and follows: Conference Rank (#1 = 10 points, #10 = 1 pooint) + conference wins - conference losses.
Oklahoma would be awarded [ ( 10 + 7 - 1) + 7 ] / 2 = 11.5 points. (The Pac-10 will have a further step of multiplying their number by 8/9 to correct for playing nine conference games when all other conferences play eight) This number will be referred to as the RCR, the Real Conference Rank.
1. Oklahoma - 11.5 points
2. Texas - 11 points
3. Texas Tech - 10.5 points
4. Oklahoma State - 7 points
5. Missourri - 6.5 points
6. Nebraska - 6 points
7. Kansas - 4 points
These numbers are alot more indicative of how the conference really should be ranked: OU, Texas and Tech close to each other at the top and the Kansas at the bottom worth a litttle more than 1/3 of the top teams.
At this point one last change is made to the rankings, they are converted to percentage points. This is done in order to even out the different number of teams each conference sends to Bowls, and does not affect the points assigned to teams according to conference rank and records.
This is done by adding together all points assigned to the teams, in this case 56.5. Then dividing each teams points by 56.5 and multiply by 100 to get the percentage. This number will be called the PCR, the percentage conference ranking.
1. Oklahoma - 20.35%
2. Texas - 19.47%
3. Texas Tech - 18.58%
4. Oklahoma State - 12.39%
5. Missourri - 11.50%
6. Nebraska - 10.62%
7. Kansas - 7.08%
Now we have the teams in the conference ranked in a way that corrects for both inter-conference issueas and intra-conferences problems. Hooray.
The next issue to tackle is the system to decide how many points the team gets to keep for its conference. This will be a bit more complicated than keeping the points of winning teams and losing the points from the beaten teams. It will take into account that an Oklahoma team that loss the national championship game by ten is still worth more than a Kansas team that blew out Minnesotta, but probably not as much as a Texas team that did win its BCS bowl.
This system will also take into account margin of victory, the percentage conference bowl record of the team they played (OCBR), the teams own PCR, the other teams RCR. The maximum amount of points awarded by the following formula is the teams own PCR, and the minimum possible value is zero. This number will be referred to as the teams BPE, Bowl Points Earned.
For Win:
PCR/2 + *{[(Opp RCR + MOV) * (OCBR)] + (Opp. RCR + MOV)] /8}
*If the second part of the equation is greater than the teams own PCR/2 it will be adjusted to equaling the Teams own PCR/2, in this case the team will retain its entire PCR as BPE.
For Loss:
PCR * MOV - [(1-OBCR) * 4] - [(1 / OPP RCR) * ]
MOV
21 or aove win +3 loss 45%
14 -20 win +2 loss 55%
7-13 win +1 loss 65%
0-6 win +.5 loss 75%
Ovetime win Even loss 85%
For Texas: (19.47/2) + {[(11.5 +.5 * .1428) + (11.5+.5)]/4} = 13.17 BPE
For Oklahoma: 20.35 *.65 - [(1-.75) *4] - [(1/11.5) *10] = 11.36 BPE
In the above cases Texas was hinderred by the combination of only beating Ohio St. by 3 (+.5)points when Ohio State came from a conference with a horrible win percentage (.1428). Thus they were awarded only 3.43 extra points for the win even though Ohio States RCR was extremely high at 11.5.
Oklahoma was not penalized too much because they lost by 10, lost to a team with an extremely high RCR at 11.5, and lost to a team with a confernce great win record of .75. Thus they still kept a little bit more than half their points.
Conference BPE results will follow as I calculate them.