|ABOUT THE RANKINGS|
While I don't plan on going into details about mathematics or statistical models, I will give a brief overview of the factors that are taken into account by the formula.
Teams are expected to win. For each win a team earns some points, but not many, especially when compared to what a team loses for each loss. Each loss takes off about 4 times as many points as a team gets for a win; loses are more important than almost any other factor. More recent games are not rated more heavily than past games.
STRENGTH OF SCHEDULE
A team earns their strength of schedule points whether they win or lose. SOS is most important as a balance to a team's loses. For example Team A loses to Team B who happens to be #1. Team A takes a 100 point hit for the loss, but gets 70 SOS points for playing the #1 team; a net loss of only 30 points. (BTW-this example is nowhere near how the actual SOS is determined and these numbers are not the actual values; they are used only to show how SOS can balance the win-loss record.)
While point margin is taken into account, it is not a major factor. Once a team has about a 3 touchdown lead, there is no further advantage to running up the score. Point margin is not a factor when teams are extremely mismatched; for example, even if #10 beats #104 by 99 points, they will get absolutely no points for it.
Expected result bonuses/penalties are given in two cases: 1) when a team that was expected to win loses, or 2) win a team that was expected to win big has a close win. Point margin and difference of ranks are used to determine whether the expected results were met. UPSET EXAMPLE: If the #5 team loses to the #10 team it's a minor upset so the bonus/penalty won't be very big. However, when #5 loses to #100 the bonus/penalty can be substantial. CLOSE CALL EXAMPLE: A good example of this is the 1999 Penn State/Pittsburgh game. The game came down to a blocked field goal at the end; the score was 20-17 for Penn State. Now the #1 team in the country should have beat the #79 team by more than 3 points! Penn State got a penalty that was almost as much as taking a loss; they fell to #4 even though they won the game. Pittsburgh took the hit for losing, but got a bonus for keeping the game close (not to mention a lot of SOS points for playing the #1 team).
I like a stable poll. Teams should not move up or down in the rankings unless something on the field warrants it. To these ends, the Whitson Rankings use a couple of other items that some other polls do not use: preseason rankings and idle placeholder points.
Preseason rankings are the same as the final previous rankings. The scores, however, are compressed. For example, at the end of 1999 there was about a 600 point difference between the first place team and the last place team. In 2000, the teams are ranked the same to begin the season, but the difference between the first place team and the last place team is only about 100 points.
When a team has an idle week they are given a few placeholder points to prevent other teams from unnecessarily jumping over them. These points are not permanent; they are removed once the team resumes play. Idle week placeholder points are not given at the end of the year for teams not playing in bowls.