Statistical modelling · College sport

Machine Learning College Predictions.
Each game predicted.

Statalysts fits a single power rating to every college football and basketball team from the season's scoring margins, anchored to a preseason projection and pulled toward each team's own results as the evidence accumulates. Those ratings produce a projected margin and win probability for every upcoming game. Want to create your own matchup? Open the custom predictor and see how the model rates any two teams.

This week's football board Open the custom predictor

Play of the day

Football Widest disagreement

UCF 54.1%

Oklahoma State 45.9%

The model makes UCF a 1.2-point favorite, which is 11.7 points clear of the market's line. This is the largest gap between the model and the market on this week's board.

Model line 1.2 pts

UCF

Market line Oklahoma State -10.5

Edge 11.7 pts

Win probability 54.1%

One game a day · the widest gap on the board See the full football board

This week

Week 6 recommended games

The games where the model disagrees most with the posted market line. A wide gap is a measure of disagreement,click any game for the full breakdown.

55 games this week

Matchup	Model line	Market line	Edge	Win prob
UCF 3-2 18 Oklahoma State 3-1 Sat, Oct 10 · 12:00 PM ET	UCF −1.2	Oklahoma State -10.5	11.7	54.1%
54.1% UCF 45.9% 18 Oklahoma State UCF54.1% Oklahoma State45.9% Projected margin UCF by 1.2 Disagreement with market 11.7 pts strong Venue Boone Pickens Stadium — Stillwater, OK
North Carolina 2-2 25 Pittsburgh 5-0 Sat, Oct 10 · 12:00 PM ET	Pittsburgh −14.6	Pittsburgh -3.5	11.1	89.0%
11.0% North Carolina 89.0% 25 Pittsburgh North Carolina11.0% Pittsburgh89.0% Projected margin Pittsburgh by 14.6 Disagreement with market 11.1 pts strong Venue Acrisure Stadium — Pittsburgh, PA
James Madison 5-0 Georgia Southern 2-3 Sat, Oct 10 · 7:30 PM ET	James Madison −18.6	James Madison −7.5	11.1	93.4%
93.4% James Madison 6.6% Georgia Southern James Madison93.4% Georgia Southern6.6% Projected margin James Madison by 18.6 Disagreement with market 11.1 pts strong Venue Allen E. Paulson Stadium — Statesboro, GA
20 Iowa 4-1 Washington 3-2 Fri, Oct 9 · 9:00 PM ET	Iowa −5.9	Washington -3.0	8.9	69.9%
69.9% 20 Iowa 30.1% Washington Iowa69.9% Washington30.1% Projected margin Iowa by 5.9 Disagreement with market 8.9 pts strong Venue Husky Stadium — Seattle, WA
9 Ole Miss 3-1 Vanderbilt 3-2 Sat, Oct 10 · 3:30 PM ET	Ole Miss −2.1	Ole Miss −9.5	7.4	57.6%
57.6% 9 Ole Miss 42.4% Vanderbilt Ole Miss57.6% Vanderbilt42.4% Projected margin Ole Miss by 2.1 Disagreement with market 7.4 pts strong Venue FirstBank Stadium — Nashville, TN
Charlotte 0-5 North Texas 3-2 Sat, Oct 10 · 3:30 PM ET	North Texas −35.3	North Texas -28.5	6.8	99.4%
0.6% Charlotte 99.4% North Texas Charlotte0.6% North Texas99.4% Projected margin North Texas by 35.3 Disagreement with market 6.8 pts strong Venue DATCU Stadium — Denton, TX

Showing the 6 widest gaps of 55 games

The full board can be filtered by conference and week, and sorted by edge or kickoff.

See all 55 games Basketball board

Last week

How the recommended games finished

Every game the model flagged on the most recently completed slate, graded on both the outright winner and the spread. Click any game for the full breakdown.

Games graded 33

Straight up 27–6 81.8%

Against the spread 16–15–2 51.6%

2 games had no posted line and are left out of the spread record.

Date	Final score	Model called	Recommended	Outcome
2026-10-04	Cincinnati 7 Arizona 34	Arizona by 2.358% win prob	Cincinnati 6.2 pt edge	Winner No cover
41.8% Cincinnati 58.2% Arizona What the model gave each side before kickoff. Cincinnati41.8% Arizona58.2% Model projected Arizona by 2.3 Actual result Arizona by 27 Market line Arizona -8.5
2026-10-04	Indiana 47 Rutgers 15	Indiana by 28.098% win prob	Indiana 4.5 pt edge	Winner Covered
98.2% Indiana 1.8% Rutgers What the model gave each side before kickoff. Indiana98.2% Rutgers1.8% Model projected Indiana by 28.0 Actual result Indiana by 32 Market line Indiana −23.5
2026-10-04	Fresno State 26 Washington State 6	Fresno State by 0.853% win prob	Fresno State 3.3 pt edge	Winner Covered
52.7% Fresno State 47.3% Washington State What the model gave each side before kickoff. Fresno State52.7% Washington State47.3% Model projected Fresno State by 0.8 Actual result Fresno State by 20 Market line Washington State -2.5
2026-10-03	Toledo 39 Ball State 24	Toledo by 26.698% win prob	Toledo 7.1 pt edge	Winner No cover
97.8% Toledo 2.2% Ball State What the model gave each side before kickoff. Toledo97.8% Ball State2.2% Model projected Toledo by 26.6 Actual result Toledo by 15 Market line Toledo −19.5
2026-10-03	Western Michigan 20 Buffalo 17	Western Michigan by 18.293% win prob	Western Michigan 4.7 pt edge	Winner No cover
93.1% Western Michigan 6.9% Buffalo What the model gave each side before kickoff. Western Michigan93.1% Buffalo6.9% Model projected Western Michigan by 18.2 Actual result Western Michigan by 3 Market line Western Michigan −13.5
2026-10-03	Miami 41 Clemson 13	Miami by 20.595% win prob	Miami 4.0 pt edge	Winner Covered
94.9% Miami 5.1% Clemson What the model gave each side before kickoff. Miami94.9% Clemson5.1% Model projected Miami by 20.5 Actual result Miami by 28 Market line Miami −16.5
2026-10-03	Texas Tech 29 Colorado 7	Texas Tech by 18.193% win prob	Texas Tech 4.6 pt edge	Winner Covered
93.0% Texas Tech 7.0% Colorado What the model gave each side before kickoff. Texas Tech93.0% Colorado7.0% Model projected Texas Tech by 18.1 Actual result Texas Tech by 22 Market line Texas Tech −13.5
2026-10-03	Texas Southern 10 Florida Atlantic 66	Florida Atlantic by 38.0100% win prob	Texas Southern 5.5 pt edge	Winner No cover
0.4% Texas Southern 99.6% Florida Atlantic What the model gave each side before kickoff. Texas Southern0.4% Florida Atlantic99.6% Model projected Florida Atlantic by 38.0 Actual result Florida Atlantic by 56 Market line Florida Atlantic -43.5
2026-10-03	Virginia 7 Florida State 38	Virginia by 7.374% win prob	Virginia 4.8 pt edge	Winner No cover
74.0% Virginia 26.0% Florida State What the model gave each side before kickoff. Virginia74.0% Florida State26.0% Model projected Virginia by 7.3 Actual result Florida State by 31 Market line Virginia −2.5
2026-10-03	UCF 17 Houston 27	Houston by 5.268% win prob	UCF 7.3 pt edge	Winner Covered
32.3% UCF 67.7% Houston What the model gave each side before kickoff. UCF32.3% Houston67.7% Model projected Houston by 5.2 Actual result Houston by 10 Market line Houston -12.5
2026-10-03	Purdue 24 Illinois 17	Illinois by 15.190% win prob	Illinois 5.1 pt edge	Winner No cover
10.3% Purdue 89.7% Illinois What the model gave each side before kickoff. Purdue10.3% Illinois89.7% Model projected Illinois by 15.1 Actual result Purdue by 7 Market line Illinois -10.0
2026-10-03	West Virginia 42 Iowa State 45	Iowa State by 10.281% win prob	Iowa State 7.2 pt edge	Winner Push
18.9% West Virginia 81.1% Iowa State What the model gave each side before kickoff. West Virginia18.9% Iowa State81.1% Model projected Iowa State by 10.2 Actual result Iowa State by 3 Market line Iowa State -3.0
2026-10-03	Ohio State 31 Iowa 14	Ohio State by 9.379% win prob	Iowa 4.7 pt edge	Winner No cover
79.2% Ohio State 20.8% Iowa What the model gave each side before kickoff. Ohio State79.2% Iowa20.8% Model projected Ohio State by 9.3 Actual result Ohio State by 17 Market line Ohio State −14.0
2026-10-03	Marshall 17 James Madison 45	James Madison by 31.999% win prob	James Madison 13.4 pt edge	Winner Covered
1.0% Marshall 99.0% James Madison What the model gave each side before kickoff. Marshall1.0% James Madison99.0% Model projected James Madison by 31.9 Actual result James Madison by 28 Market line James Madison -18.5
2026-10-03	Ohio 13 Kent State 10	Ohio by 11.484% win prob	Ohio 8.4 pt edge	Winner Push
83.6% Ohio 16.4% Kent State What the model gave each side before kickoff. Ohio83.6% Kent State16.4% Model projected Ohio by 11.4 Actual result Ohio by 3 Market line Ohio −3.0
2026-10-03	McNeese 14 LSU 63	LSU by 57.5100% win prob	LSU 4.0 pt edge	Winner No cover
0.0% McNeese 100.0% LSU What the model gave each side before kickoff. McNeese0.0% LSU100.0% Model projected LSU by 57.5 Actual result LSU by 49 Market line LSU -53.5
2026-10-03	Eastern Michigan 38 Massachusetts 14	Massachusetts by 12.385% win prob	Massachusetts 6.3 pt edge	Winner No cover
14.7% Eastern Michigan 85.3% Massachusetts What the model gave each side before kickoff. Eastern Michigan14.7% Massachusetts85.3% Model projected Massachusetts by 12.3 Actual result Eastern Michigan by 24 Market line Massachusetts -6.0
2026-10-03	Bowling Green 24 Miami (OH) 20	Miami (OH) by 20.395% win prob	Miami (OH) 6.8 pt edge	Winner No cover
5.2% Bowling Green 94.8% Miami (OH) What the model gave each side before kickoff. Bowling Green5.2% Miami (OH)94.8% Model projected Miami (OH) by 20.3 Actual result Bowling Green by 4 Market line Miami (OH) -13.5
2026-10-03	Alabama 56 Mississippi State 23	Alabama by 9.178% win prob	Alabama 3.1 pt edge	Winner Covered
78.5% Alabama 21.5% Mississippi State What the model gave each side before kickoff. Alabama78.5% Mississippi State21.5% Model projected Alabama by 9.1 Actual result Alabama by 33 Market line Alabama −6.0
2026-10-03	Florida 17 Missouri 45	Florida by 9.379% win prob	Florida 4.8 pt edge	Winner No cover
79.1% Florida 20.9% Missouri What the model gave each side before kickoff. Florida79.1% Missouri20.9% Model projected Florida by 9.3 Actual result Missouri by 28 Market line Florida −4.5
2026-10-03	Louisville 28 NC State 31	NC State by 0.251% win prob	NC State 5.7 pt edge	Winner Covered
49.4% Louisville 50.6% NC State What the model gave each side before kickoff. Louisville49.4% NC State50.6% Model projected NC State by 0.2 Actual result NC State by 3 Market line Louisville −5.5
2026-10-03	Maryland 23 Nebraska 48	Nebraska by 18.894% win prob	Nebraska 4.3 pt edge	Winner Covered
6.4% Maryland 93.6% Nebraska What the model gave each side before kickoff. Maryland6.4% Nebraska93.6% Model projected Nebraska by 18.8 Actual result Nebraska by 25 Market line Nebraska -14.5
2026-10-03	UTEP 7 New Mexico 61	New Mexico by 32.899% win prob	New Mexico 10.3 pt edge	Winner Covered
0.9% UTEP 99.1% New Mexico What the model gave each side before kickoff. UTEP0.9% New Mexico99.1% Model projected New Mexico by 32.8 Actual result New Mexico by 54 Market line New Mexico -22.5
2026-10-03	Notre Dame 37 North Carolina 26	Notre Dame by 27.498% win prob	Notre Dame 6.3 pt edge	Winner No cover
98.0% Notre Dame 2.0% North Carolina What the model gave each side before kickoff. Notre Dame98.0% North Carolina2.0% Model projected Notre Dame by 27.4 Actual result Notre Dame by 11 Market line Notre Dame −21.0
2026-10-03	Wyoming 0 North Dakota State 28	North Dakota State by 24.397% win prob	North Dakota State 5.8 pt edge	Winner Covered
3.0% Wyoming 97.0% North Dakota State What the model gave each side before kickoff. Wyoming3.0% North Dakota State97.0% Model projected North Dakota State by 24.3 Actual result North Dakota State by 28 Market line North Dakota State -18.5
2026-10-03	UTSA 16 Rice 14	UTSA by 15.490% win prob	UTSA 3.9 pt edge	Winner No cover
90.0% UTSA 10.0% Rice What the model gave each side before kickoff. UTSA90.0% Rice10.0% Model projected UTSA by 15.4 Actual result UTSA by 2 Market line UTSA −11.5
2026-10-03	UL Monroe 35 South Alabama 52	South Alabama by 20.795% win prob	South Alabama 6.7 pt edge	Winner Covered
4.9% UL Monroe 95.1% South Alabama What the model gave each side before kickoff. UL Monroe4.9% South Alabama95.1% Model projected South Alabama by 20.7 Actual result South Alabama by 17 Market line South Alabama -14.0
2026-10-03	Temple 17 South Florida 13	South Florida by 10.281% win prob	South Florida 4.2 pt edge	Winner No cover
19.0% Temple 81.0% South Florida What the model gave each side before kickoff. Temple19.0% South Florida81.0% Model projected South Florida by 10.2 Actual result Temple by 4 Market line South Florida -6.0
2026-10-03	Samford 14 UAB 33	UAB by 23.396% win prob	Samford 7.2 pt edge	Winner Covered
3.5% Samford 96.5% UAB What the model gave each side before kickoff. Samford3.5% UAB96.5% Model projected UAB by 23.3 Actual result UAB by 19 Market line UAB -30.5
2026-10-03	Syracuse 42 UConn 41	Syracuse by 0.452% win prob	UConn 6.1 pt edge	Winner Covered
51.6% Syracuse 48.4% UConn What the model gave each side before kickoff. Syracuse51.6% UConn48.4% Model projected Syracuse by 0.4 Actual result Syracuse by 1 Market line Syracuse −6.5
2026-10-03	Michigan State 3 Wisconsin 31	Wisconsin by 6.271% win prob	Michigan State 4.2 pt edge	Winner No cover
29.0% Michigan State 71.0% Wisconsin What the model gave each side before kickoff. Michigan State29.0% Wisconsin71.0% Model projected Wisconsin by 6.2 Actual result Wisconsin by 28 Market line Wisconsin -10.5
2026-10-02	North Texas 45 Tulsa 44	North Texas by 6.973% win prob	North Texas 8.4 pt edge	Winner Covered
72.8% North Texas 27.2% Tulsa What the model gave each side before kickoff. North Texas72.8% Tulsa27.2% Model projected North Texas by 6.9 Actual result North Texas by 1 Market line Tulsa -1.5
2026-10-02	Pittsburgh 35 Virginia Tech 33	Pittsburgh by 5.970% win prob	Pittsburgh 9.4 pt edge	Winner Covered
70.0% Pittsburgh 30.0% Virginia Tech What the model gave each side before kickoff. Pittsburgh70.0% Virginia Tech30.0% Model projected Pittsburgh by 5.9 Actual result Pittsburgh by 2 Market line Virginia Tech -3.5

Power ratings

Top ten, both sports

A rating is a team's strength in points. The difference between two ratings is the model's projected margin on a neutral floor.

#TeamRating spreadRating

Alabama 30.6

Ohio State 29.8

Notre Dame 29.5

Georgia 27.7

Miami 27.7

Indiana 27.4

Texas 25.5

Utah 24.3

Oregon 24.1

LSU 22.3

Full top 25 for both sports

Every rated team, ordered by rating, updated as results come in.

Open power rankings

Method

How a rating becomes a prediction

Collect the season

Every completed game is pulled from the college football and basketball data APIs: final scores, schedule, venue and, where posted, the market line. Football rates FBS and FCS teams on the same scale, because a large share of FBS games are played against FCS opponents and those results only mean something if both sides carry a real rating.

Fit a rating to each team

A ridge regression fits one rating per team to the season's scoring margins, so a team's number reflects who it played as well as by how much it won. Early in the season there is little to fit on, so the regression is pulled toward a preseason projection built from the team's own final rating last season and a published preseason rating.

As games accumulate, the weight on this season's results grows and the prior fades.

Project the game

The difference between two ratings, plus home advantage where it applies, is the projected margin. That margin converts to a win probability for each side.

Where a sportsbook has posted a line, the model's number is compared against it. The gap between the two is the edge — not a claim that the model is right, but a measure of where it disagrees most with the market.