Tulane vs Duke Prediction, Picks & Odds — Sep 5, 2026

Duke grades 7.6 points better than an 11-3 Tulane team, and 4.5 of that is the game being in Durham. The next-largest piece is the portal, worth about two points of the margin, and it runs one direction: Tulane sent 27 players out and brought 20 in, a net loss of 21 stars. That is a teardown rather than churn, and it is why a team that won 11 games is laying more than a touchdown in September.

0%

-38

1%

-32

2%

-24

5%

-18

9%

-10

13%

-4

17%

+4

18%

+10

15%

+18

11%

+24

6%

+32

3%

+38

1%

+46

Tulane winsDuke wins

Market: Duke −9Projected: Duke by 8 (69%)

The range of outcomes for this game. Each bar is how often the final margin lands in that range — the more it leans toward Duke, the more likely Duke wins.

GRIDPEX.COM

The one edge Tulane owns is the one it can no longer staff

Duke's pass defense ranked 125th of 136 last season, the bottom ten percent of the sport, and Tulane's passing game ranked 26th. That is the largest unit edge on the board at 0.6 standard deviations, and it belongs to the underdog. Tulane returns 14 percent of the passing production that earned the rank, so the group getting credit for that number is largely playing somewhere else this fall.

Anthony Brown produced +0.36 EPA per play on a 16 percent target share, and Zycarl Lewis Jr. was better per snap at +0.52 but on only 8 percent of the targets, which projects to 11 to 36 receiving yards here. Destyn Hill arrives from LSU as the one four-star addition to that group. Duke's answer to a 125th-ranked secondary was two portal corners, Evan Smith from Northwestern and Dylan Flowers from Western Kentucky.

All of it runs through Zeon Chriss-Gremillion, a Houston transfer with no season of record and no FBS snaps. Asking an unseen passer to attack a bottom-ten secondary downfield is the hardest version of this assignment, and Tulane's 26th-ranked passing grade describes players who are gone. Whether the three who stayed can reproduce it in Durham is the widest error bar in this game.

Duke's receiving corps is a rebuild too; its quarterback is not

Duke ranked 24th throwing it and returns 12 percent of that production. Darian Mensah, Que'Sean Brown and Cooper Barkate, three four-stars, all left, two of them to Miami. Walker Eget arrives from San José State with four FBS seasons and +0.220 career EPA per play, which makes him the only projected starter in this game with a real body of work, even though none of it happened in this offense.

That pushes Duke toward the ground, where 43 percent of last season's rushing production returns. Nate Sheppard carried a 45 percent rush share and projects for 57 to 109 yards with a 55 percent chance of a touchdown, and he runs into the better of the two run defenses, with Tulane favored by 0.2 standard deviations there. Jeremiah Hasley, 17 to 54 receiving yards, is the early-down outlet Eget can reach without testing a corner.

Tulane's pass rush lost its edge; Duke's front never finished plays

Tulane's defense was the better unit last season by disruption: 34 sacks (17th), 15 interceptions (12th), 81 tackles for loss (19th). Three producers behind that are gone, edge Harvey Dyson to NC State, four-star corner Jahiem Johnson to Arkansas and Jayden Lewis to Alabama A&M. Chris Rodgers (2.0 sacks, 75 tackles), Kevin Adams III (two interceptions) and E'zaiah Shine (seven pass breakups) are the returning spine, with Kajuan Banks in from South Florida to patch the corner.

Duke's front inverts that profile: 87 tackles for loss ranked ninth nationally while 26 sacks ranked 58th, a defense that lives in the backfield without finishing at the quarterback. Chriss-Gremillion's inexperience hands Luke Mergott (80 tackles) and DaShawn Stone (77 tackles, two interceptions) a cheaper route to the same result, first-down tackles for loss that turn Tulane's downfield plan into third-and-long, where a 125th-ranked secondary becomes far easier to hide.

Duke's margin over the nine sits in the turnover column

Duke covers nine points in 48 percent of projected outcomes, and the split behind that is clean: Duke wins the turnover battle by two or more in 21 percent of games, while 66 percent come out even in giveaways and stay low-event. Winning the giveaway battle wins roughly three of every four games at any level, so the claim is not that turnovers matter; it is that Duke's cover has almost no other source.

Tulane's upset is the clean, low-event version

The path is no giveaways, Jamauri McClure grinding toward the top of a 33 to 92 yard band behind a run game that returns 60 percent of its production, and two vertical shots landing on that 125th-ranked secondary. Tulane allowed 23.9 points a game last season while Duke allowed 29.4, and the projection gives a one-score finish a 38 percent chance. That is the version where Duke's nine points never matter.

Tulane's incoming class ranks 92nd with no blue-chips, so the 20 portal arrivals are the replacement plan rather than a supplement to it. Duke's class ranks 124th, but its portal work came out plus-15 on position-weighted stars, which is how a team that lost its starting quarterback and both four-star receivers still projects as the sturdier roster. Neither side is deep; Tulane is the one asking more new names to be right in week one.

Sheppard's carries and two imported corners decide this

Duke needs Sheppard to hold up against the tougher run defense so Eget's four seasons get spent on second-and-five rather than third-and-nine, and it needs Smith and Flowers to survive alone against Brown and Hill. Tulane needs a first-time starter to be functional rather than good, and needs the rush to travel without Dyson. Keep the turnover column level and nine points is more than this game is likely to produce.