I do not wish to give out all of the secrets of the formula but I can offer these insights.
Each season the beginning ratings are adjusted from the previous year by using each team's schedule to determine strength of schedule. Added in are factors for experience (lettermen and starters returning). Everything is purely numerical. There is nothing given consideration to one team that isn't apllied to all 117, and no changes have been made to my system since the end of Year #2.
Each week, the teams are adjusted based on actual outcome of the game vs. the computer's prediction.
Over- or under-performance adjusts the team's rating. Wild fluctuations are avoided by utilizing a season-long determining factor.
Unlike some polls which qualify a rating for non Division 1-A teams, my ranking recognizes as those teams as 0.00. So teams that pad their records with fluff opponents are penalized severely.
My contention is, and it used to be a determing factor in bowl selections, teams should not be able to use wins over 1-AA or lower teams to pad their record.
By the way, I am anti-playoff. I like the bowl system and feel a playoff would kill many football programs at the numerous schools that have no real chance in ever winning a national title. And attendance would suffer immediately at those schools.
Additionally, a playoff would not draw fan support. Fans can not afford to travel to 3 different sites to follow their teams through the playoffs. So the vast majority would wait to see if their team made it to the title game and then maybe buy tickets, if they could get them, if they could afford them, and if they had enough time to make travel plans.
Dave Congrove / dcongrove at cfl dot rr dot com