The history of the CFP: Sometime in the middle of the 2005 college football season, I began participating in a college pick'em game on a website sponsored by some Auburn fans. For the first few weeks, I worked out some statistics by hand - namely, the average points forced and allowed by each team. By comparing a team's offense to their opponent's defense (and vice versa), I hoped to correctly predict the winners.

Unsurprisingly, this method was unsuccessful, due in large part to the simplicity of my formula. Over the next few weeks, I gathered more and more score data and revised my formula until I reached something that produced reasonable results. Eventually, I started compiling matchups into a large Excel spreadsheet, which generated constants close to the ones used in my formula now.

By the end of the season, I had formatted the original spreadsheet to a very readable design and had begun backtracking my formula to previous years. I was hooked on the challenge of correctly predicting college football winners, and a new hobby was born.

Changes to the system: In 2006, I eliminated the results of matches against I-AA teams. In the off-season, I searched for a correlation between previous year's ending rating, the number of returning starters, and current year's ending rating. No such correlation was found, either on offense or defense. However, I decided to begin the 2007 season with the ratings from 2006. These preseason ratings will be phased out exponentially as current year scores become available. This year's CFP will run from Monday through Sunday so that I have more time to prepare upcoming weeks.

This season, I may look into eliminating overtime points from the score totals.