Division I-A College Football

Explanation of Solecismic Software Ratings

The ratings system is based on an assessment of each game played between two Division I-A college football teams.

The final score matters, but since the game score is curved using a "normal" distribution and there's a limit to how high a game score a team can achieve, running up the score doesn't make much difference.

A home-field advantage is built into the system, and a small emphasis is placed on having a halftime lead.

Once the game scores have been adjusted, I set up a grid of all 119 Division I-A teams and compare each team to every other team, passing through as many relevant game scores as possible. The algorithm will use as many as seven steps to make a comparison between two teams. Shorter links, with the shortest being an actual game between the two teams, are given precedence over longer links.

I then rate the teams based on the percentage of favorable comparisons between teams. A perfect score of 100 means that team has a positive game score in comparison to all 118 possible Division I-A opponents.

For the first six weeks of the season, last year's rating, modified by the number of returning starters, is used as a diminishing portion of each team's rating.