How is a 3 and 2 cal team ranked 9th? also a va tech team ranked 18th who lost to a fcs school. and a good fbs school team. also a 3-3 arizona state team all ranked above iowa?
It is the way the formula works. Basically, if Team A has played team C, and team C has played team B, and team A has played team B:
Team B beats C and team A, team C beats team A = rankings of Team B, C, A.
So common opponents matter a ton in his system. VT playing an FCS school and losing doesn't hurt them SOS wise, because that FCS school hasn't lost to any other FBS teams and therefore cannot statistically weigh on the SOS component of his ranking.
Also, his ranking tries to weigh recent losses more heavily than distant past losses: if Team A beats Team B this week, but lost to Team C the first week, and Team B beat Team C the second week, the rankings will go 1)A, 2)C, 3)B because, in his words, teams should not be ranked higher than the team they were beat by this week.
Over a full season, the number of common opponent nodes or related opponent nodes (i.e. Iowa State being a common opponent of Texas Tech and Iowa for example) grows to the point where you can statistically predict how games would go between any two teams in FBS, as well as take into account "who's hot" right now. At the beginning of the season, there is less data so the predictor gets more and more accurate the number of games played. That's why the Sagarin rankings don't mean much until late in the season, when they solidify and are more difficult to change. This early, there will be big swings in the Sagarin rankings as teams and their common opponents meet and win/lose. In one ranking, the margin of victory matters. In the other it doesn't. The two rankings often paint a very different picture of College Football.
