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

Pittsburgh 93.3%

Boston College 6.7%

The model makes Pittsburgh a 18.5-point favorite, which is 12.0 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 18.5 pts

Pittsburgh

Market line Pittsburgh −6.5

Edge 12.0 pts

Win probability 93.3%

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

This week

Week 7 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.

62 games this week

Matchup	Model line	Market line	Edge	Win prob
Pittsburgh 5-1 Boston College 2-3 Sat, Oct 17 · 12:00 PM ET	Pittsburgh −18.5	Pittsburgh −6.5	12.0	93.3%
93.3% Pittsburgh 6.7% Boston College Pittsburgh93.3% Boston College6.7% Projected margin Pittsburgh by 18.5 Disagreement with market 12.0 pts strong Venue Alumni Stadium (Chestnut Hill, MA) — Chestnut Hill, MA
New Mexico 4-1 Hawai'i 1-5 Sat, Oct 17 · 10:30 PM ET	New Mexico −25.3	New Mexico −14.0	11.3	97.4%
97.4% New Mexico 2.6% Hawai'i New Mexico97.4% Hawai'i2.6% Projected margin New Mexico by 25.3 Disagreement with market 11.3 pts strong Venue Clarence T.C. Ching Athletics Complex — Honolulu, HI
11 Utah 5-0 Colorado 2-3 Sat, Oct 17 · 10:15 PM ET	Utah −24.7	Utah −15.5	9.2	97.1%
97.1% 11 Utah 2.9% Colorado Utah97.1% Colorado2.9% Projected margin Utah by 24.7 Disagreement with market 9.2 pts strong Venue Folsom Field — Boulder, CO
Louisiana Tech 3-2 UL Monroe 0-5 Sat, Oct 17 · 8:00 PM ET	Louisiana Tech −23.7	Louisiana Tech −15.5	8.2	96.7%
96.7% Louisiana Tech 3.3% UL Monroe Louisiana Tech96.7% UL Monroe3.3% Projected margin Louisiana Tech by 23.7 Disagreement with market 8.2 pts strong Venue Malone Stadium — Monroe, LA
Virginia 4-2 24 SMU 4-1 Sat, Oct 17 · 12:00 PM ET	SMU −2.3	SMU -9.5	7.2	58.3%
41.7% Virginia 58.3% 24 SMU Virginia41.7% SMU58.3% Projected margin SMU by 2.3 Disagreement with market 7.2 pts strong Venue Gerald J. Ford Stadium — University Park, TX
18 Oklahoma State 4-1 20 Houston 5-1 Sat, Oct 17 · 7:30 PM ET	Houston −11.6	Houston -4.5	7.1	83.9%
16.1% 18 Oklahoma State 83.9% 20 Houston Oklahoma State16.1% Houston83.9% Projected margin Houston by 11.6 Disagreement with market 7.1 pts strong Venue TDECU Stadium — Houston, TX

Showing the 6 widest gaps of 62 games

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

See all 62 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 27

Straight up 18–9 66.7%

Against the spread 13–12–2 52.0%

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

Date	Final score	Model called	Recommended	Outcome
2026-10-11	Kansas 24 Utah 41	Utah by 22.096% win prob	Utah 6.5 pt edge	Winner Covered
4.1% Kansas 95.9% Utah What the model gave each side before kickoff. Kansas4.1% Utah95.9% Model projected Utah by 22.0 Actual result Utah by 17 Market line Utah -15.5
2026-10-10	Eastern Michigan 25 Akron 23	Eastern Michigan by 3.362% win prob	Akron 3.7 pt edge	Winner Covered
61.6% Eastern Michigan 38.4% Akron What the model gave each side before kickoff. Eastern Michigan61.6% Akron38.4% Model projected Eastern Michigan by 3.3 Actual result Eastern Michigan by 2 Market line Eastern Michigan −7.0
2026-10-10	Georgia 20 Alabama 17	Alabama by 5.368% win prob	Alabama 7.8 pt edge	Winner No cover
31.8% Georgia 68.2% Alabama What the model gave each side before kickoff. Georgia31.8% Alabama68.2% Model projected Alabama by 5.3 Actual result Georgia by 3 Market line Georgia −2.5
2026-10-10	Tulane 17 Army 20	Army by 8.978% win prob	Army 5.9 pt edge	Winner Push
21.9% Tulane 78.1% Army What the model gave each side before kickoff. Tulane21.9% Army78.1% Model projected Army by 8.9 Actual result Army by 3 Market line Army -3.0
2026-10-10	Iowa State 10 BYU 24	BYU by 10.181% win prob	Iowa State 4.5 pt edge	Winner Covered
19.2% Iowa State 80.8% BYU What the model gave each side before kickoff. Iowa State19.2% BYU80.8% Model projected BYU by 10.1 Actual result BYU by 14 Market line BYU -14.5
2026-10-10	South Carolina 38 Florida 19	Florida by 11.383% win prob	South Carolina 3.2 pt edge	Winner Covered
16.6% South Carolina 83.4% Florida What the model gave each side before kickoff. South Carolina16.6% Florida83.4% Model projected Florida by 11.3 Actual result South Carolina by 19 Market line Florida -14.5
2026-10-10	James Madison 27 Georgia Southern 24	James Madison by 18.693% win prob	James Madison 11.1 pt edge	Winner No cover
93.4% James Madison 6.6% Georgia Southern What the model gave each side before kickoff. James Madison93.4% Georgia Southern6.6% Model projected James Madison by 18.6 Actual result James Madison by 3 Market line James Madison −7.5
2026-10-10	UAB 13 Memphis 38	Memphis by 20.595% win prob	Memphis 6.5 pt edge	Winner Covered
5.1% UAB 94.9% Memphis What the model gave each side before kickoff. UAB5.1% Memphis94.9% Model projected Memphis by 20.5 Actual result Memphis by 25 Market line Memphis -14.0
2026-10-10	Tulsa 26 Navy 34	Tulsa by 0.853% win prob	Tulsa 3.3 pt edge	Winner No cover
53.0% Tulsa 47.0% Navy What the model gave each side before kickoff. Tulsa53.0% Navy47.0% Model projected Tulsa by 0.8 Actual result Navy by 8 Market line Navy -2.5
2026-10-10	Charlotte 10 North Texas 50	North Texas by 35.399% win prob	North Texas 6.8 pt edge	Winner Covered
0.6% Charlotte 99.4% North Texas What the model gave each side before kickoff. Charlotte0.6% North Texas99.4% Model projected North Texas by 35.3 Actual result North Texas by 40 Market line North Texas -28.5
2026-10-10	Air Force 28 Northern Illinois 7	Air Force by 12.085% win prob	Air Force 4.5 pt edge	Winner Covered
84.7% Air Force 15.3% Northern Illinois What the model gave each side before kickoff. Air Force84.7% Northern Illinois15.3% Model projected Air Force by 12.0 Actual result Air Force by 21 Market line Air Force −7.5
2026-10-10	Ball State 14 Northwestern 45	Northwestern by 40.5100% win prob	Northwestern 4.0 pt edge	Winner No cover
0.3% Ball State 99.7% Northwestern What the model gave each side before kickoff. Ball State0.3% Northwestern99.7% Model projected Northwestern by 40.5 Actual result Northwestern by 31 Market line Northwestern -36.5
2026-10-10	Stanford 7 Notre Dame 58	Notre Dame by 43.6100% win prob	Notre Dame 5.2 pt edge	Winner Covered
0.2% Stanford 99.8% Notre Dame What the model gave each side before kickoff. Stanford0.2% Notre Dame99.8% Model projected Notre Dame by 43.6 Actual result Notre Dame by 51 Market line Notre Dame -38.5
2026-10-10	UCF 28 Oklahoma State 42	UCF by 1.254% win prob	UCF 10.7 pt edge	Winner No cover
54.1% UCF 45.9% Oklahoma State What the model gave each side before kickoff. UCF54.1% Oklahoma State45.9% Model projected UCF by 1.2 Actual result Oklahoma State by 14 Market line Oklahoma State -9.5
2026-10-10	San Diego State 14 Oregon State 66	Oregon State by 11.383% win prob	San Diego State 3.2 pt edge	Winner No cover
16.6% San Diego State 83.4% Oregon State What the model gave each side before kickoff. San Diego State16.6% Oregon State83.4% Model projected Oregon State by 11.3 Actual result Oregon State by 52 Market line Oregon State -14.5
2026-10-10	North Carolina 21 Pittsburgh 17	Pittsburgh by 14.789% win prob	Pittsburgh 11.1 pt edge	Winner No cover
11.0% North Carolina 89.0% Pittsburgh What the model gave each side before kickoff. North Carolina11.0% Pittsburgh89.0% Model projected Pittsburgh by 14.7 Actual result North Carolina by 4 Market line Pittsburgh -3.5
2026-10-10	Buffalo 23 Toledo 8	Toledo by 28.098% win prob	Toledo 7.0 pt edge	Winner No cover
1.8% Buffalo 98.2% Toledo What the model gave each side before kickoff. Buffalo1.8% Toledo98.2% Model projected Toledo by 28.0 Actual result Buffalo by 15 Market line Toledo -21.0
2026-10-10	North Dakota State 7 UNLV 13	North Dakota State by 7.173% win prob	North Dakota State 4.1 pt edge	Winner No cover
73.4% North Dakota State 26.6% UNLV What the model gave each side before kickoff. North Dakota State73.4% UNLV26.6% Model projected North Dakota State by 7.1 Actual result UNLV by 6 Market line North Dakota State −3.0
2026-10-10	Washington State 16 Utah State 17	Utah State by 2.258% win prob	Washington State 3.3 pt edge	Winner Covered
42.2% Washington State 57.8% Utah State What the model gave each side before kickoff. Washington State42.2% Utah State57.8% Model projected Utah State by 2.2 Actual result Utah State by 1 Market line Utah State -5.5
2026-10-10	Ole Miss 52 Vanderbilt 17	Ole Miss by 2.158% win prob	Vanderbilt 7.9 pt edge	Winner No cover
57.5% Ole Miss 42.5% Vanderbilt What the model gave each side before kickoff. Ole Miss57.5% Vanderbilt42.5% Model projected Ole Miss by 2.1 Actual result Ole Miss by 35 Market line Ole Miss −10.0
2026-10-10	Syracuse 25 Virginia 28	Virginia by 12.986% win prob	Virginia 3.5 pt edge	Winner No cover
13.6% Syracuse 86.4% Virginia What the model gave each side before kickoff. Syracuse13.6% Virginia86.4% Model projected Virginia by 12.9 Actual result Virginia by 3 Market line Virginia -9.5
2026-10-10	Iowa 41 Washington 24	Iowa by 5.970% win prob	Iowa 7.4 pt edge	Winner Covered
70.0% Iowa 30.0% Washington What the model gave each side before kickoff. Iowa70.0% Washington30.0% Model projected Iowa by 5.9 Actual result Iowa by 17 Market line Washington -1.5
2026-10-10	Arizona 52 West Virginia 31	Arizona by 6.572% win prob	Arizona 3.5 pt edge	Winner Covered
71.6% Arizona 28.4% West Virginia What the model gave each side before kickoff. Arizona71.6% West Virginia28.4% Model projected Arizona by 6.5 Actual result Arizona by 21 Market line Arizona −3.0
2026-10-10	Kent State 7 Western Michigan 38	Western Michigan by 19.994% win prob	Western Michigan 6.4 pt edge	Winner Covered
5.5% Kent State 94.5% Western Michigan What the model gave each side before kickoff. Kent State5.5% Western Michigan94.5% Model projected Western Michigan by 19.9 Actual result Western Michigan by 31 Market line Western Michigan -13.5
2026-10-08	South Alabama 56 Arkansas State 49	Arkansas State by 4.967% win prob	Arkansas State 3.4 pt edge	Winner No cover
33.2% South Alabama 66.8% Arkansas State What the model gave each side before kickoff. South Alabama33.2% Arkansas State66.8% Model projected Arkansas State by 4.9 Actual result South Alabama by 7 Market line Arkansas State -1.5
2026-10-08	South Florida 24 UTSA 31	UTSA by 1.154% win prob	South Florida 5.9 pt edge	Winner Push
46.0% South Florida 54.0% UTSA What the model gave each side before kickoff. South Florida46.0% UTSA54.0% Model projected UTSA by 1.1 Actual result UTSA by 7 Market line UTSA -7.0
2026-10-07	Jacksonville State 27 Kennesaw State 26	Kennesaw State by 1.455% win prob	Kennesaw State 3.9 pt edge	Winner Covered
45.1% Jacksonville State 54.9% Kennesaw State What the model gave each side before kickoff. Jacksonville State45.1% Kennesaw State54.9% Model projected Kennesaw State by 1.4 Actual result Jacksonville State by 1 Market line Jacksonville State −2.5

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

Notre Dame 31.0

Alabama 29.8

Ohio State 29.1

Miami 28.6

Georgia 28.1

Indiana 27.2

Utah 23.9

Texas 23.9

Oregon 22.8

LSU 21.6

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.