A new set of college football rankings for us to play with!

That feeling is in the air… it’s college football time again, and with it comes the return of all-out obsessive coverage on Da Blog. Both lineal titles (college and NFL) have been belatedly updated, including the new 2009 Boise State title and Super Bowl XLIV title. (I’ll have a post on the new holder of 2006 Boise State coming soon.) Although my Da Blog Poll came out to two votes to keep the College Football Schedule to one to junk it, I’m getting rid of it anyway. I need all the free time I can get to work on other things, and along with the College Football Rankings, starting Week 3 I’ll be premiering a new college football concept that has a lot more reason to premiere at the point any two teams can be connected to one another through a series of games… and one that could prove to be a lot more time-consuming than the Schedule ever was.

I started thinking about this with regards to combat sports like boxing and MMA, which I may extend this concept to eventually. If any sport has a more confusing title situation than college football, it’s those two (and horse racing), with all the different weight classes, not to mention all the different sanctioning bodies in the former. But for all the confusion over who the champ is, how the champ is determined is fairly straightforward: to be the man, you have to beat the man. So long as the champion does not lose, that person will remain the champion. This is taken to the point where lists of rankings will actually separate out the champion from the ranked fighters. No matter how strong a record you may rack up, to be the man, you have to beat the man. The championship system in combat sports is predicated on the notion that the result of a single fight is representative of which fighter is better overall. The same principle should be in play for ranking fighters below the champion.

Now, in what other sport is this the case? I don’t just ask this rhetorical question because I already created the college football lineal title on the same notion. You regularly hear the argument that Team A is better than Team B because Team A beat Team B, even if it was by one point in overtime at home. In a sense, this is the philosophy behind the BCS Title Game, as well as, to a lesser extent, the Super Bowl. (In most other sports a series of games determines the champion, removing some of the uncertainty and ambiguity of a single game.) You take what you think is the top two teams, pit them against each other, and the winner is the champion, as well as considered “better”. ¬†As I pointed out last year, 2005 USC may well have been as good as ESPN said they were when they infamously started comparing the Trojans to all the great teams of the past, but we take it as given that Texas was the better team, because they beat USC. And BCS arguments are regularly settled by comparing whether one of the teams under discussion beat the other.

So I’m introducing what I call the line-of-sight rankings, to bring if not objectivity, at least consistency to the criteria we already use to argue about college football. Every team is situated below all the teams it lost to and above all the teams it beat. Obviously, there will be contradictions in the rankings, and in those cases we’ll have to throw out some games. We’ll determine what games to throw out in this order:

  • If two or more different contradictions can be resolved by throwing out a single game, throw out that game. Throw out the game that resolves the most contradictions, except that if a game is the most recent game for at least one team, it is considered to resolve one fewer contradiction than it actually does.
  • Otherwise, always eliminate home-team victories before neutral-site games, and neutral-site games before road-team victories.
  • Among games of similar siting, for every full 10 points of the margin of victory, add one to the week number. Then eliminate the game with the lowest week number, but do not eliminate a team’s most recent game. In event of a tie, eliminate the game with the smaller margin of victory. If there is still a tie, add the total number of losses for the winning team to the total number of wins by the losing team, and eliminate the game where that number is higher. If there is still a tie, remove the prohibition on eliminating a team’s most recent game, and if that does not help, subtract the losing team’s C Rating from the winning team’s C Rating, and eliminate the game where that number is lower.

Because every team doesn’t play every other team in college football, there will still be ambiguity in the rankings. If a team’s worst relevant loss is to the #5 team, and their best relevant win is to the #10 team, where between those two numbers is the team itself ranked? I settle these situations as follows:

  • If there is a “pod” of only one team as described above, including undefeated teams, rank the team directly ahead of the best team beaten in a relevant win. Winless teams are ranked directly behind their worst relevant loss. The team in question will have the rank of their worst relevant loss in parenthesis or, if undefeated in relevant games but not #1, have their entry boldfaced.
  • If there are two or more “pods” of multiple teams each that can be ranked a certain way between any two teams (or at the top or bottom of the rankings), or if there are two individual teams that can be ranked between another two teams but whose ranking vis-a-vis one another is unclear, break them up and rank them separately, within their own pods. Each team’s rank is listed as their best possible ranking except at the top of the rankings, when it is their worst possible ranking. In the case of the individual teams, they are listed as tied and in C Rating order unless one has a lineal title.

I’ll whip out the first rankings Week 3, when they become meaningful, and we’ll see how they play themselves out over the course of the season, and how much work they add to my already heavy workload.