description
John Coffey's Football Ratings
My system was developed as a hobby and out of the love of the game. The formulation of the model is complex and based on the concept that winning / losing is important, but scoring margins should also be considered. Since the ratings are based on a blend of "predictive" and "retrodictive" algorithms, they have a tendency to show a high correlation to the consensus of a group of models.
The ratings are unbiased and the basic components include:
Relative scoring margins (margins of a team compared with its opponent's margins)
Winning percentage
Opponent's winning percentages
Strength of schedule which is based of quality of opponents, opponents' opponents, etc.
The programming is advanced and repeatedly recalculates the ratings until they stabilize, and automatically considers the strength of opposition to an extended depth.
Since the model does not use information from previous seasons, the rankings become more meaningful and less variable as the season progresses. The ratings represent a team's relative scoring potential at a neutral site. To facilitate comparisons, the highest rated team is assigned a rating of 100. For college football, predictions based on the power ratings can be obtained by comparing two team's ratings and applying a home field advantage. I believe that 3.0 points is a good general estimate for home field advantage for college football.
For the NFL ratings, a more involved prediction model is used to provide the actual score predictions. This prediction model includes calculations for home field advantage trends, away field disadvantage trends, and momentum (time sensitive weighting).
Questions or comments may be directed to coffeyratings@gmail.com
Disclaimer: The contents of this website are provided for informational/entertainment purposes only. The owner of this website does not encourage or endorse the use of this content for any other purposes, and under no circumstances will be held responsible for damages related to its content.