DATA AND RATINGS EXPLANATION

BASIS OF RATINGS SYSTEM:

*Season Progression: Later games are worth more than earlier games (i.e. the last game is worth about 1.5 times the first game)

*Strength of Schedule: Measures a team's opponent's record and their opponent's opponents combined records and is factored against the performance points earned (see below).

*Game Location: Location of where the game is played factors into scoring system (home, away, neutral). Every team is given the same factor for location since it is very difficult to accurately quantify home field advantage.

*Overall Record: Reflects only wins over 1-A opponents but losses include all games played. A small bonus is awarded for an undefeated season. There is a small penalty assesed for each loss.

*Performance Points: Based on how a team fares against opponent A compared to how all other teams perform against team A. Both offensive performance (points scored vs the opponent's avg points allowed) and defensive performance (points allowed vs the opponent's average points scored) are measured. Bonus points awarded is capped to avoid too many points for running up the score.

*Non 1-A games: Games against lower division opponents account for very minimal points in the total unless a team loses to a lower division team. A penalty is assesed for losing to a non 1-A team, and a small penalty is factored in for each non 1-A opponent played.

*What is not factored: No other polls, either human or computer affect these ratings. Prior year's data also does not affect these ratings, so therefore there is no preseason poll (the first poll will be released around the 3rd or 4th week of the season). All teams start with the same amount of points at the beginning of the year regardless of the prior year's points or any other factors.

*Notes: In the beginning of the season, the ratings are very volatile due to limited data. The first poll will be released after the games of September 25th, but these will still be based on somewhat limited data. I expect things to normalize after games of October 9th. One last note...the system is bias free. In other words, every team is measured with the same factors (won-loss record, SOS, bonus points, etc) counting equally.

THE CONFERENCE RANKINGS:

*Simple average versus weighted average: The simple average is the total combined points of all teams within a conference divided by the number of teams in the conference. The weighted average uses a scale in this fashion:

8 team conference 1-2-3-4-4-3-2-1

9 team conference 1-2-3-4-5-4-3-2-1

10 team conference 1-2-3-4-5-5-4-3-2-1

11 team conference 1-2-3-4-5-6-5-4-3-2-1

12 team conference 1-2-3-4-5-6-6-5-4-3-2-1

14 team conference 1-2-3-4-5-6-7-7-6-5-4-3-2-1

This means that a conference's top team points are multiplied by 1, the 2nd team by 2, the 3rd team by 3, and so on. The sum of the weighted scores is then divided by the number of weighted scores to determine the weighted average. This is also known as the central mean.

NOTE: The closer the weighted average is to the simple average, the more balanced a conference is. The bigger the difference, the more a conference is top heavy or bottom heavy. If the weighted average is less than the simple average, it indicates a top heavy conference, if the weighted average is more than the simple average it indicates a bottom heavy conference.

HOW DOES THIS COMPARE TO OTHER COMPUTER POLLS?

Visit Kenneth Massey's College Football Ranking Comparison site at http://www.mratings.com/cf/compare.htm (link at bottom of page) to view a compilation of 80 national computer polls. I ran a correlation to the Massey consensus 1-117 ratings and came back with a 0.986 for 2002 which would rank it tied for 24th out of 80 for these polls. For the correlation to the Massey consensus 1-25 ratings, it came back with a 0.954 for 2002, ranking 29th out of 80 for these polls. Here are the correlation scores for the 1997-2003 polls versus the Massey consensus rankings:

1997 0.986 (Tied-5th of 47polls)

1998 0.987 (Tied-9th of 64 polls)

1999 0.983 (Tied-20th of 66 polls)

2000 0.983 (Tied-26th of 64 polls)

2001 0.985 (Tied-18th of 72 polls)

2002 0.986 (Tied-24th of 80 polls)

2003 0.988 (Tied-31st of 97 polls)

On Massey's site, he lists the polls with the highest correlations to the consensus on the left side. As you can see, this rating system consistently falls between 0.983 and 0.988 against the consensus, and therefore has a high degree of reliability.

Also, please be sure to check out David Wilson's American College Football site which includes a list of computer rating systems.

Special thanks to Kenneth Massey and David Wilson for including this site on their pages!

If you have any questions, comments, or suggestions please email me at gunner596@aol.com

IN THE WORKS:

*Prior year ratings (1991 season just finished and I will work backwards through the years as time allows) to be posted as time allows. Links will be posted on the homepage as each is entered on the site.

*All-time teams (as more prior seasons are rated).