Talisman Red's College Football Computer Ratings

The Wilson Formula

One of the first rating systems I ever saw the source code for was one by Mr. David Wilson (Wilson Performance Ratings; UW-Milwaukee '65, '66, UW-Madison '67). I loved its simplicity, and I have adapted it for numerous sports over the last 20 years. David passed away in January 2024 at the age of 80, and he left an important mark on the early days of the college football internet (see more in the About page). I am happy to continue producing his ratings to honor that legacy.

Wilson's system does *not* produce predictions, point spreads, etc.; it only serves as a way to rate teams based on games played. All games are weighted evenly (although his code can be set up to count postseason games twice -- I'll do that here too), and game locations and points scored/allowed do not matter.

Brief Methodology

I'm working on writing up a more detailed description of David's formula (his website was scrubbed from the UW servers a few years ago, but god bless the Internet Archive). In the meantime, it is pretty much this:

"For each game a team plays it gets a Game Performance Rating. This is equal to the opponent's rating plus 100 if the team won or minus 100 if the team lost. The team's Performance Rating is the average of its Game Performance Ratings. All of the Game Performance Ratings are recalculated each week based on the latest ratings for the opponents."

The bottom line: if you win against strong teams, you get a higher rating. If you lose against weaker teams, you get a lower rating. Sum up the results, and that's it.

Something new this year: Using the Wilson system to rate conferences based on their non-conference games. Here are the D 1 results, treating all teams in a conference as a single team.

--- Division I FBS ---
                        W  L  T
                        -  -  -
  1  Miami (FL)         8  0     950
  2  Utah               8  0     947
     UCLA               8  0     947
  4  Indiana            8  1     938
  5  Kentucky           7  1     936
  6  Texas              6  2     923
  7  Georgia            7  1     921
  8  Alabama            6  2     910
     Mississippi        6  2     910
 10  Wake Forest        7  1     894
     Ohio State         6  2     894
 12  Louisiana St.      6  2     892
 13  Florida            5  3     891
 14  Notre Dame         7  1     890
 15  South Carolina     5  3     888
     Iowa               6  2     888
 17  Texas Tech         7  1     887
 18  Brigham Young      7  1     885
 19  Tennessee          6  2     876
 20  Houston            6  2     875
 21  Arizona            6  2     871
     Mississippi St.    4  4     871
 23  Oregon             5  3     869
 24  Minnesota          5  3     866
 25  Oklahoma           4  4     865

--- Division I FCS ---
                        W  L  T
                        -  -  -
  1  Montana St.        8  1     772
  2  Stephen F.Austin   9  0     761
  3  N. Arizona         7  2     760
  4  Youngstown St.     7  2     757
  5  S. Dakota St.      7  2     745
  6  Tarleton St.       9  0     739
  7  Tennessee Tech     8  1     724
  8  Harvard            6  1     720
  9  South Dakota       7  3     717
 10  Lamar              7  2     712
     Montana            6  3     712
 12  Monmouth (NJ)      7  2     709
 13  Austin Peay        7  2     704
 14  Illinois St.       7  2     700
 15  Campbell           7  2     695
 16  Princeton          5  2     692
 17  North Dakota       6  4     691
 18  Weber St.          6  4     689
 19  Abilene Christian  5  4     688
 20  Idaho              4  5     687
 21  Incarnate Word     5  4     682
 22  E. Tennessee St.   6  3     681
     Idaho St.          6  3     681
 24  N. Iowa            4  4     677
 25  Albany (NY)        6  3     676

Input files

• names.txt (List of team names, going back to 1800s)
• games.txt (Games played - in 2025, collected by Dr. Peter Wolfe)
• conf.txt (Conference affiliations)

Output files

• byrating.txt (Teams listed in order by rating)
• byname.txt (Teams listed alphabetically)
• byrate1.txt (Division I FBS by rating)
• byrate2.txt (Division I FCS by rating)
• byrate3.txt (Division II by rating)
• byrate4.txt (Division III by rating)
• byrate5.txt (NAIA by rating)
• standing1.txt (Division I FBS Conference standings)
• standing2.txt (Division I FCS Conference standings)
• standing3.txt (Division II Conference standings)
• standing4.txt (Division III Conference standings)
• standing5.txt (NAIA Conference standings)
• gam_neat.txt (games.txt made neat)

An adjustment I make

David's philosophy was that every team started the season with the same rating, regardless of who they were. I disagree a little bit, and so early in the season, I use games from last year to provide a more meaningful starting point for the ratings. By the time everyone has played about 8 games, that influence is gone and the ratings are "true" to the current season.

Source code

I have saved David's original source code from 1994 in a Github repository. Feel free to download it, run it for college football or another sport or even your own league, and tell me how it goes.

More detailed methodology

[I'll add this soon...]