Acknowledgements: I could not have reached the point I am at today without access to the work of those who have gone before me. For example, James Howell's historical score database allowed me to test my prediction system on the past few years of college football. Howell's page is hosted by David Wilson, whose yearly lists of schedules for all 120 Division I-A teams make programming those schedules into my spreadsheets a lot easier. And, of course, much of my system's structure was taken from the ranking systems listed on this ranking comparison page, courtesy of Kenneth Massey. I thank everyone who made their own systems available online for newcomers like me to view.

Explanation: Currently, the CFP is a large Excel file that I set up before every college football season. I manually program the schedules of all 119 Division I-A college football teams before the season begins. During the season, I manually enter the scores of all the games played. Needless to say, these are both very time-consuming tasks. Perhaps one day I will write some sort of automated program to make this hobby a little easier.

My formula currently takes the following statistics into account when generating a team's rating: a) the average points forced and allowed by the team and b) the average points forced and allowed by the team's opponents. To predict the winner of a game, I compare the two teams' ratings and give the home team some advantage points.

Teams like Louisville, Texas Tech, and Boise State often rank very high in listings of all the teams from best rating to worst. This is because of their tendency to score as many points as time will allow, as well as the disparity among the other teams in their conferences or conference divisions. Because it produces results which are rather illogical (Texas Tech was ranked #3 at the end of 2005; Virginia Tech was ranked well above Florida State, despite losing the ACC Championship), I do not use the CFP to generate rankings. I have only found it useful in predicting the straight-up winners of college football games.

The goal: A 100% straight-up prediction rate. Just kidding. Right now, I'm shooting for a correct prediction rate of 75%.