Welcome to my humble webpage. My name is Gary and if you're a college football fan like I am, then you've probably had some complaints about the current polls in place (AP, Coach's, BCS, Harris Interactive, etc). We all do. That is why I chose to do some research into developing a ranking model that's not only fair and objective, but uses the retrodictive philosophy.
I am a senior mathematics major at Stetson University in Central Florida. I am attempting to extend a model developed by Dr. Jay Coleman of the University of North Florida. His model directly addresses and minimizes what are known as game score violations (When one team defeats another yet the defeated team is ranked higher in the rankings). His model (MinV) is mixed-integer binary programming program and so is mine except with some non-linear qualities. You can see the results of his work for previous seasons here.
I have so far attempted to incorporate strength of schedule, average points per game, average points per game allowed, wins, and losses. I am not directly addressing home-field advantage nor margin of victory at this time. This still requires more research of the current model.
All my game data was obtained from CBS Sportsline. Other data, such as strength of schedule is calculated by me using the current BCS formula: SOS = (2/3)* Opponent's Winning% + (1/3)* Opponent's Opponent's Winning%. I also take into account a separate SOS component for the strength of teams beaten. All wins are wins against FBS(formally 1A) opponents only. Only losses against FCS(formally Div1AA) will count towards a team's record.
*Special thanks to Kenneth Massey for adding my rankings to his comparison page.
For questions, comments, or suggestions contact me here
