Power Ranking the FBS

One thing that MLB and the NBA have going for them is the length of their seasons.  Their scheduling methodology allows each team to play a preponderance of other teams in their conference, and there are only two conferences.  This allows analysts to compare most teams directly when trying to evaluate one against another.

College football, for all its awesomeness, doesn't even come close to allowing that.  FBS teams play in 10 conferences with a few independents playing games across several conferences.  The 3 or 4 non-conference games that an FBS team can fit into its annual schedule are often planned more for ensuring a victory (or a fat paycheck for smaller schools) or filling concessionaire coffers than for creating opportunities to evaluate football skill.

The College Football Playoff, by considering strength of schedule, may change this somewhat as teams now have a greater incentive to play more challenging games.  But this is really more symbolism over substance.  There simply aren't enough Saturdays available to ensure that teams play enough football to make reasonably objective assessments of who is better than who.

With 128 FBS teams, each would have to play approximately 40 randomly assigned games to have a 90% chance of being no more than 1 game removed from every other team. For instance.  Ohio State lost to Virginia Tech which lost to East Carolina.  Ohio State is one game removed from East Carolina in this scenario.

Clearly, a 40 game, computer-generated random schedule is never going to happen.  So we have to rely on other, much more subjective means to rack and stack teams for the college football playoff.

That presents the obvious question of "What is important and what isn't?"  The College Football Playoff has already said that strength of schedule will be considered in the committee's choices.  In the BCS, margin of victory was not considered in the computer polls, even though it is highly correlated to overall winning percentage.

I believe that a computer ranking model that indexes margin of victory by strength of schedule presents the most viable ranking system.  It is a methodology that fine tunes itself as the season progresses.  It is also the most defensible.  Nothing is more highly correlated to winning that scoring offense and scoring defense.

My ranking model uses scoring offense, scoring defense, season-to-date strength of schedule, and a small adjustment for winning percentage to capture other intangibles in the game.  Using this to make all teams play a model 127 game, round robin schedule we can get a good picture is who is playing better than who at this point in the season.

Through three games, the top 10 and most likely College Football Playoff contenders are:

There is little separation between the top 5.  In fact, the point separation between 1 to 5 is the same as between 5 and 6 and 6 to 10.  Note the conspicuous absence of the Big Ten from the Top Ten. The average rankings of each conference make the degree to which the Big Ten has slipped into gravitated toward mediocrity is clear.  The SEC's ranking at the top by a large margin, is even more impressive when one considers that Vanderbilt is dead last in the FBS.

For the morbidly curious and Big Ten fans, here's the rest of the FBS.