A.I. College Football Picks for Week 12 – ATS & O/U Predictions from Sportradar Model

• SportRadar’s college football AI went 4-6 both ATS and on totals in Week 11, dropping the model further below .500 as it continues trending in the wrong direction
• The model correctly backed Texas A&M to blow out Missouri and nailed Iowa covering against Oregon, but whiffed on Navy getting 27.5 points in a 39-point loss to Notre Dame
• Below, the AI’s Week 12 picks lean heavily on road underdogs and unders, including fading USC as a road favorite and backing Pittsburgh to keep it close against Notre Dame

SportRadar’s AI model continues sliding backward, posting another losing 4-6 record both ATS and on totals in Week 11. That’s now 48-51-1 ATS on the season, with the model hemorrhaging value at an accelerating pace. Anyone tailing these picks is down serious money after accounting for the juice.

The college football AI picks had some bright spots, like correctly calling Texas A&M to demolish Missouri 38-17 and backing Iowa to stay within the number at Oregon. But missing badly on BYU getting blown out by Texas Tech (when the model had BYU +10.5) and Navy losing by 39 when getting 27.5 points shows the algorithm still can’t identify basic mismatches.

Week 12 features several rivalry games and playoff elimination matchups. The AI has studied the college football odds and emerged with a conservative slate dominated by unders and road dogs.

AI College Football Picks for Week 12

AI picks record to date:

• ATS picks: 48-51-1
• O/U picks: 52-48

College football AI predictions generated on November 14, 2025. Check out the top college football betting apps ahead of Week 12.

Unders Everywhere You Look

Nine of ten games lean under, with only Virginia-Duke getting the over treatment at 58.5. The model apparently thinks November college football turns into a defensive slog across the board. Michigan-Northwestern under 41.5 at Wrigley Field makes sense given the venue and weather concerns, but some of these other unders feel lazy.

Alabama-Oklahoma under 45.5 stands out as particularly aggressive. These are two proud programs with playoff implications on the line. Alabama’s offense has been explosive all season while Oklahoma’s defense has shown vulnerabilities. The model sees a defensive grind, but this could easily sail over if either team gets rolling early.

The lone over pick on Duke-Virginia looks contrarian given Duke’s recent defensive improvements. But the model thinks both offenses find success in what could be a back-and-forth ACC shootout.

Road Dogs and Questionable Favorites

Iowa getting 6.5 at USC actually makes sense. The Hawkeyes’ defense travels well and USC has been inconsistent at home. Kirk Ferentz teams always show up in big spots, and the Coliseum won’t be as intimidating for a veteran Iowa squad.

Pittsburgh getting 12.5 against Notre Dame feels like the model chasing points. The Irish just destroyed Navy 49-10 and have been rolling. Pitt’s won seven straight but hasn’t faced anyone near Notre Dame’s caliber during that streak. This could get ugly if the Irish start fast.

Navy getting 10 points at home against South Florida looks tempting after the Bulls have been inconsistent on the road. But Navy just got embarrassed by Notre Dame and might be demoralized. The model’s backing the triple-option to keep it close, which rarely works out when facing athletic defenses.

Chalk Plays That Make Sense

Alabama laying 5.5 to Oklahoma feels right given the Sooners’ road struggles this season. Georgia covering 6.5 against Texas also tracks with the Bulldogs’ home dominance and Texas’s offensive line issues. Cincinnati laying 6.5 to Arizona at home should hit given the Bearcats’ playoff positioning and home-field advantage.

Utah as a small road favorite at Baylor makes sense too. The Utes have been road warriors all season while Baylor’s struggled with consistency. BYU laying 3.5 to TCU feels tight, but the Cougars’ home-field advantage in Provo could be the difference.

AI College Football Picks – Results from Week 11

Another losing week pushes the model deeper into the red. Going 4-6 ATS and 4-6 on totals means anyone betting $100 per game lost $260 on spreads and another $260 on totals after accounting for juice. That’s over $500 gone in a single week.

The few bright spots included nailing the Iowa-Oregon under at 34 total points and correctly backing Wake Forest to cover against Virginia. But massive misses like thinking Navy could stay within 27.5 points of Notre Dame (they lost by 39) and backing Indiana to cover 14.5 when they only won by 3 show the model has no feel for momentum or matchups.

At 48-51-1 ATS and 52-48 on totals, the AI is basically a coin flip on spreads while clinging to a tiny edge on totals that’s rapidly evaporating. Following these picks has been a one-way ticket to the poorhouse.