CFB

Dan’s Computer Ratings and How They Work

If you followed our college football podcast in 2021, you might remember regular references to my computer. I’ve now been running a computer algorithm for the past five college football seasons, and I thought this would be a good time to explain my ratings and how they work.

Basically, the computer will pick every game involving two FBS schools, or games involving one FBS school and either Sam Houston State, Jacksonville State or North Dakota State. The way that it makes its pick is that it first rounds each team’s rating to the nearest whole number, then subtracts the lower rating from the higher rating. It then adds three points for home-field advantage to create its final pick. For example, if Minnesota is scheduled to play at Penn State, the computer would currently expect the Nittany Lions to win by eight points.

There is no limit to how high or low a team’s number can get. Generally, only the best teams get above 60, while most good teams will sit between 40 and 50. At the other end of the spectrum, teams in single digits or negative numbers would struggle to beat quality FCS opponents.

HOW DO RATINGS CHANGE?

After the game, I enter the adjusted final score into the computer’s spreadsheet, giving the visiting team an extra three points — unless the visiting team was shut out. This is important, because the computer has a provision that rewards good defense. Winning a game 23-6 is more valuable to the computer than winning by 34-17, even though it’s the same point difference. Because of this, if the visiting team is shut out, I instead subtract three points from the home team. Using the Minnesota-Penn State example again, if Penn State beats Minnesota 23-10 in Happy Valley, the computer reads it as a 23-13 win for Penn State.

Based on the adjusted score, what week it is and the strength of the two teams, the computer spits out a new rating for each team. Generally, if a team beats the computer’s expectations, its rating improves. If it falls short, its rating decreases. There are rare exceptions, as the computer tempers its expectations when one team is favored by 40 or more. This is because it wouldn’t be fair to expect Georgia to keep pouring it on to maintain its rating when it faces the likes of Kent State.

The computer also looks at the whole season as a picture, not just one game. This is why games decrease in importance as the season progresses; the computer adds each game to its complete collection of data.

WHAT ARE THE LEAGUES?

I run an unofficial competition in my ratings where teams are grouped into geographic leagues, as my own stand against the idiocy that is USC and UCLA moving to the Big Ten. The leagues are my ideal system, which would have the four division winners face off for the national championship. Like soccer, these leagues include promotion and relegation. At the end of the season, the worst two teams in each division in League A will be relegated to League B, while the best two will be promoted to League A. Promotions and relegations will continue down the line through all of Division I.

HOW ACCURATE IS IT?

The computer tends to pick the winner of the game 75% of the time. If you’re looking for a moneyline parlay, this might be a good place to start. On the spread, it’s closer to 55%.

Here are the ratings after Week 1. (Note: Florida State-LSU and Clemson-Georgia Tech have not yet played.)

LEAGUE A EAST

Ohio State 61.15

Michigan 58.33

Penn State 47.78

Cincinnati 45.55

Kentucky 43.48

Michigan State 42.41

Pittsburgh 39.74

Wake Forest 39.54

N.C. State 38.88

Appalachian State 34.33

Louisville 25.29

 

LEAGUE A SOUTHEAST

Georgia 72.65

Alabama 65.42

Tennessee 49.05

Clemson 42.45

Auburn 42

Mississippi State 41.24

Mississippi 40.3

Florida 39.94

Houston 36.73

Louisiana 35.67

LSU 34.75

 

LEAGUE A MIDWEST

Notre Dame 50.88

Iowa State 47.8

Oklahoma State 47.41

Arkansas 46.93

Wisconsin 46.01

Oklahoma 44.23

Minnesota 42.88

Purdue 40.47

Iowa 40.31

Kansas State 40.2

Nebraska 33.6

 

LEAGUE A WEST

Texas A&M 49.4

Baylor 46.24

Utah 44.61

Texas 41.92

BYU 40.86

Arizona State 39.59

UCLA 37.94

Washington State 36.72

Air Force 36.07

Oregon 33.75

Boise State 33.7

 

LEAGUE B EAST

Syracuse 37.51

West Virginia 34.93

Central Michigan 30.77

Maryland 30.38

Marshall 28.42

Army 27.56

Rutgers 27

Toledo 25.12

Boston College 24.81

Western Michigan 23.27

Miami-Ohio 23.23

 

LEAGUE B SOUTHEAST

North Carolina 36.6

Florida State 34.95

Miami 34.48

Coastal Carolina 34.21

East Carolina 33.01

South Carolina 32.78

Virginia 32.01

Central Florida 29.62

Virginia Tech 24.19

Georgia Tech 22.13

Liberty 22.11

 

LEAGUE B MIDWEST

SMU 43.58

TCU 36.78

Western Kentucky 35.63

Texas Tech 33.75

Illinois 32.66

UTSA 30.57

UAB 28.85

Missouri 27.89

Tulane 23.61

Tulsa 22.67

Memphis 21

 

LEAGUE B WEST

Oregon State 41.64

USC 37.72

Washington 34.21

Fresno State 33.44

California 32.18

Nevada 29.6

Stanford 25.11

San Diego State 20.87

Colorado State 20.18

Utah State 17.88

Colorado 16.93

 

LEAGUE C EAST

Eastern Michigan 18.44

Duke 17.45

Ohio 16.53

Kent State 14.43

Navy 13.62

Buffalo 12

Bowling Green 9.15

Connecticut 6.58

Akron -2.4

Massachusetts -5.57

Temple -7.03

 

LEAGUE C SOUTHEAST

James Madison 34.38

Florida Atlantic 21.37

Old Dominion 20.93

Georgia State 20.85

Troy 19.34

South Florida 13.31

South Alabama 16.1

Georgia Southern 15.41

Southern Mississippi 10.68

Charlotte 0.19

Florida International -3.66

 

LEAGUE C MIDWEST

Vanderbilt 34.2

North Dakota State 32

Northwestern 29.17

Indiana 27.79

Northern Illinois 17.47

Kansas 14.29

Ball State 11.34

Middle Tennessee 10.76

Arkansas State 10.69

Louisiana Tech 7.19

Louisiana-Monroe 6.97

 

LEAGUE C WEST

Arizona 31.05

Wyoming 18.91

UNLV 17.2

San Jose State 16.9

North Texas 13.28

UTEP 12.5

Texas State 9.22

New Mexico 8.58

New Mexico State 6.45

Hawaii 0.2

Rice -2.48

Click to comment

Leave a Reply

To Top
%d bloggers like this: