NEW RESEARCH— Every college football team, FBS to DIII, ranked on one scale

DSA LABS

support@dsa-labs.com

917-657-6883

234 Brookside Road, Darien, CT 06820

Privacy Policy
Our TeamSolutionsPredictionsLearn MoreContact Us

© 2026 DSA Labs. All rights reserved.

Veteran Owned and Operated

DSA Laboratories
Home
Solutions
Predictions
Resources
GO TO STATLINK
All track records
All divisionsFBSFCS

DSA Track Record · College Football 2026 · All divisions

How often the model gets it right

Every number here comes from win probabilities saved before kickoff and graded after the final whistle. Nothing is recalculated with today's ratings, which would make the model look better than it is.

Season record
642–252
Right–wrong, favorite to win
Accuracy
71.8%
894 games graded
Brier score
0.184
Lower is better. A coin flip scores 0.250.

LiveLive: predictions saved before kickoff, through week 11Updated Nov 16, 2026

01

Calibration: does 70% mean 70%?

For each confidence band, how often the favorite actually won. A 70–80% favorite should win about 75% of the time. Points on the dashed line are perfectly calibrated; below the line means the model was overconfident, above means it was too cautious. Bigger points mean more games.
BandGamesPredictedActualGap
50-60%25754.2%55.6%+1.4 pts
60-70%23364.9%65.7%+0.8 pts
70-80%19074.9%81.0%+6.2 pts
80-90%135

06

Per-game lookup

Pick a team to see each game's DSA win probability as it was saved before kickoff, the result, and whether the model's favorite won.
BacktestUpdated Nov 16, 2026

Backtest (rebuilt ratings, not live predictions)

Ratings rebuilt week by week for 2026 through week 11 using only games played before each week, then used to predict that week. It shows how the method would have done; it is not a record of predictions published at the time, and it is never added to the live numbers.

Season record
1,434–706
Right–wrong, favorite to win
Accuracy
67.0%
2,140 games graded
Brier score
0.207
Lower is better. A coin flip scores 0.250.

B1

Calibration: does 70% mean 70%?

For each confidence band, how often the favorite actually won. A 70–80% favorite should win about 75% of the time. Points on the dashed line are perfectly calibrated; below the line means the model was overconfident, above means it was too cautious. Bigger points mean more games.

Only predictions with a save time before kickoff are counted. Accuracy is rounded down to one decimal. The favorite is whichever team DSA gave the higher win probability. Model index-ratio-1.0.

Schedule and lines: CollegeFootballData.com

84.9%
85.2%
+0.3 pts
90-100%7993.8%97.5%+3.7 pts

02

Week by week

Right and wrong calls for every week tracked so far. Games with an exact 50/50 prediction count as neither.
WeekRight–wrongAccuracyRight versus wrong
8149–6768.9%
9168–5376.0%
10165–6472.0%
11160–6870.1%

03

By matchup type

Cross-division games are where a single all-divisions scale gets tested hardest, and they are also the easiest to call. Same-division games are the real test.
  • FBS vs FBS

    64.7%

    238 games

  • FCS vs FCS

    67.2%

    232 games

  • III vs III

    81.7%

    219 games

  • II vs II

    74.3%

    203 games

  • FBS vs FCS

    100.0%

    1 games · small sample

  • III vs NAN

    100.0%

    1 games · small sample

04

Versus the betting market

How often each side's favorite won, on the same 242 games that had a line. The market prices in injury, lineup, and weather news; the model only knows past results.

DSA favorite won

64.4%

Market favorite won

68.1%

05

Biggest misses

The games where the model was most confident and wrong. Shown in full, not buried.
WeekDSA favoritePre-game win %Lost toFinal
8Cal Poly Mustangs90.4%Utah Tech Trailblazers43-26
8Edinboro Fighting Scots90.1%Gannon Golden Knights31-14
10Sacred Heart Pioneers90.0%Towson Tigers43-40
10Wheeling Cardinals89.0%Glenville State Pioneers38-34
11Minnesota Duluth Bulldogs89.0%Minot State Beavers45-31
8Troy Trojans88.5%UL Monroe Warhawks21-14
9James Madison Dukes87.5%Troy Trojans24-21
11Rhode Island Rams87.5%Towson Tigers19-16
10Edward Waters Tigers87.4%Allen Yellow Jackets31-28
9Tiffin Dragons87.2%Lake Erie Storm21-14
BandGamesPredictedActualGap
50-60%72855.2%54.9%-0.3 pts
60-70%61564.9%63.7%-1.2 pts
70-80%44474.7%76.1%+1.5 pts
80-90%25984.3%83.8%-0.5 pts
90-100%9493.5%92.5%-0.9 pts

B2

Week by week

Right and wrong calls for every week of the backtest. Games with an exact 50/50 prediction count as neither.
WeekRight–wrongAccuracyRight versus wrong
2118–6365.1%
3128–6466.6%
4141–6468.7%
5142–7565.4%
6124–7263.2%
7139–8462.3%
8148–7765.7%
9160–6770.4%
10167–6871.0%
11167–7269.8%

B3

By matchup type

Cross-division games are where a single all-divisions scale gets tested hardest, and they are also the easiest to call. Same-division games are the real test.
  • FBS vs FBS

    62.9%

    578 games

  • FCS vs FCS

    64.2%

    500 games

  • II vs II

    72.2%

    497 games

  • III vs III

    68.7%

    473 games

  • FBS vs FCS

    74.6%

    75 games

  • FCS vs II

    61.5%

    13 games · small sample

  • III vs NAN

    33.3%

    3 games · small sample

  • FCS vs III

    0.0%

    1 games · small sample

B4

Versus the betting market

How often each side's favorite won, on the same 1,153 games that had a line. The market prices in injury, lineup, and weather news; the model only knows past results.

DSA favorite won

64.2%

Market favorite won

73.8%

B5

Biggest misses

The games where the model was most confident and wrong. Shown in full, not buried.
WeekDSA favoritePre-game win %Lost toFinal
10Wheeling Cardinals93.7%Glenville State Pioneers38-34
5Newberry Wolves93.4%UVA Wise Cavaliers48-38
5William Jewell Cardinals93.1%Indianapolis Greyhounds45-10
10Edward Waters Tigers92.4%Allen Yellow Jackets31-28
5South Carolina State Bulldogs92.3%The Citadel Bulldogs34-31
8Edinboro Fighting Scots90.6%Gannon Golden Knights31-14
10Sacred Heart Pioneers90.3%Towson Tigers43-40
8Cal Poly Mustangs89.8%Utah Tech Trailblazers43-26
8Troy Trojans89.7%UL Monroe Warhawks21-14
5Montana State Bobcats89.1%Idaho Vandals34-35