How the rankings are factored:
1. Each week, I come up with a W/L record for each conference. This becomes a baseline stat in two ways, as a Win% and a Loss%.
2. If a team wins a game, they receive points equal to the opposing conferences' win percentage.
3. If a team loses a game, they receive negative points equal to the opposing conferences' loss percentage.
4. I add up all the games. Each conference receives points based on what each team did during a given week. I then divide by the number of out of conference games that a conference played.
5. Once the points are tallied, I divide each conferences' points by the highest conferences' point total, so that the highest conference receives a score of 1.00.
Michigan beat WMU this week. The MAC had a win percentage of 0.615, so Michigan gained 0.615 points for the Big Ten. The Big Ten had a LOSS percentage of 0.167. This means that the MAC received -0.167 points for losing to Michigan. Had they lost to a lesser conference, the loss percentage of that conference would have been higher, so the point hit would have been more substantial.
FCS wins are not counted towards a conferences' points. FCS losses, however, are counted, and are very damaging. The FCS loss percentage is 0.947, which would be about as negative as losing to the Big Ten five times.
There are a few flaws in this system. One, individual teams are not accounted for. USC beating Minnesota means as much as if they beat Ohio State. This ends up balancing out, for the most part, as a larger sample size typically evens out the playing field. Two, margin of victory is not factored. While Auburn barely escaped, it counts as if they won by 4 touchdowns. I may address this in the future, but for now, it is way too much effort.