North Carolina vs. Duke Prediction, Odds & Pick
Prediction, odds and the model's read
Market vs our model
we are 9.9 toward 
DUKE
| Market | Our model | |
|---|---|---|
| Spread | DUKE −4.5 | DUKE −14.4 |
| DUKE win | 66% | 83% |
Calibrated against results, not against the closing line. Read this as our confidence, not as an edge over the market.
Scouting report
2026 projection · last season, roster and staff
Neither side projects a clear way in. North Carolina and Duke rate within touching distance on every unit matchup here, closest in the push up front — on this evidence it comes down to the things a season profile cannot see.
The two teams
North Carolina
North Carolina is in year 2 under Bill Belichick. The way in is 130th at running it, and they are 122nd at hitting chunk plays.
North Carolina has played 1 game against an average slate, adjusted for here. That football carries about 0% of what is below; the rest is still the preseason projection.
Projected soft spots
- Offenserunning it#130projected
- Offensehitting chunk plays#122projected
- Offensefirst and second down#117projected
Duke
Duke throws it: Manny Diaz's offenses have been pass-leaning across 5 seasons. The strengths are defensive, led by 24th at stuffing runs at the line. The way in is 123rd at keeping everything in front of them, and they are 119th at closing out passing downs.
On fourth down he goes for it about as often as his peers do from the same spots (+2.5%). Against what the expected-points model recommends he is −22.8% — the profession as a whole is negative on this.
Duke has played 2 games against an average slate, adjusted for here. That football carries about 0% of what is below; the rest is still the preseason projection.
Projected strengths
- Defensegetting stopped at the line#24projected
Projected soft spots
- Defensehitting chunk plays#123projected
- Defensesurviving second and twelve#119projected
- Offensegetting stopped at the line#117projected
Ranks are national. Nothing here is last season's number — carrying one forward is measurably worse than saying nothing — and in-season figures are adjusted for who was actually played. This describes the matchup; it is not a prediction and not a pick.
Head coaches only — there is no free feed for coordinators, so on a program where the head coach does not call the offense this describes the staff he hired. Play-calling identity is a style, not a grade: throwing more than the situations warrant is not better than throwing less than they warrant.
Model prediction
how the number is builtOur model simulates the game drive by drive from each side's opponent-adjusted efficiency (DUKE Elo 1615, UNC Elo 1398 for reference), with home-field advantage. That projects DUKE -14.4 (83% to win) — 9.9 points clear of DUKE's market line of -4.5. That is a disagreement, not a betting edge — sides we favour have not covered at better than breakeven.
Why that percentage is worth reading: across 604 graded in-season games, the calls this model put near 58% came in at 58.2%. The full calibration table is published, bin by bin. Early in the season the cold-start path runs instead and calibrates less well (Brier 0.2028 against 0.1889).
Model as of Sep 1 · through week 1 · drive simulation
Where the game is won
2026 national rank · right is better
offensethe defense it faces
Moving the ball
UNC41st77thDUKEDUKE85th33rdUNCThrough the air
UNC59th101stDUKEDUKE110th61stUNCOn the ground
UNC98th35thDUKEDUKE42nd48thUNCStaying on the field
UNC81st50thDUKEDUKE97th112thUNCTaking care of it
UNC68th53rdDUKEDUKE31st30thUNC
Each bar runs from last in the country on the left to 1st on the right. The filled span is the gap between the offense and the defense it faces.
Season form — 2026
nationally rankedThe matchup, in context
series history + adjusted profilesNorth Carolina leads 63–38–4 · Duke has won 2 straight · 51% have been one-possession games
| UNC | DUKE | |
|---|---|---|
| +5.5 (#50) | CORE overall | +4.6 (#53) |
| -1 / -7 | offense / defense | -1 / -5 |
| 1.44 | points / drive | 1.42 |
| 0.89 | allowed / drive | 0.25 |
| 22% | three-and-outs | 33% |
| +11 | pass over expected | -15 |
Bold is the side ahead. CORE strips the situation from every play and solves the schedule out across the league, so these compare two teams that never met.
Key matchups
The analyst read: each team's offense splits crossed against the opponent's defense, with the edges called out.
Create a free account →You get Pro free for 6 months. No card required.
FAQ
Who will win North Carolina vs. Duke?
Gridpex's model favors Duke with a 83% win probability.
What's the predicted spread for North Carolina vs. Duke?
The model projects Duke by 14.4.
What time is North Carolina vs. Duke and what channel is it on?
Time TBD, at Wallace Wade Stadium.
Where this number comes from. Every projection on this page is produced by the same model, run before kickoff on opponent-adjusted efficiency, power ratings and situational data — walk-forward validated on eleven seasons and never adjusted after the fact. How the model works →