Parent DirectoryI have developed a system of ranking division 1-a college football teams called the Craig Index. It is a simple system, and I couldn't help but notice that nearly all of the polls on this rating comparison are quite complicated. It incorporates the result of the game, opponents' winning percentage, opponents opponents' winning percentage, conference strength, and margin of victory (limit of 29).
The conference strength is calculated by dividing a conferences' team into 3rds based on conference win fraction. In an out-of-conference game the computer evaluates which 3rd the two teams fall into. If say a team in the bottom third of the Big Ten beat an SEC team in the top third the Big Ten conference would be awarded a greater margin and the SEC conference less than if a the team was in the top 3rd of the Big Ten, and its opponent in the bottom third of the SEC.
My system is retrodictive. I use no preseason results and I don't start my ratings until all teams have played 5 games. I don't know much about rating systems, but I think that most retrodictive systems weight games played later on in the season more heavily. Mine doesn't.
Daniel Craig / Buckeye319@aol.com