Duke

Profile vs. FBS, 2026

Farther out is better on every spoke. Dashed ring: the FBS median.

• EPA per play

  Expected points added on an average snap: how much each play moved the scoring odds.
  Offense+0.2781st
  Defense+0.0660th
• Success rate

  Share of snaps that stayed on schedule: half the yards needed on 1st down, 70% on 2nd, all of them on 3rd or 4th.
  Offense43.8%53rd
  Defense39.3%62nd
• Explosiveness

  Average EPA on the successful snaps: how big the good plays were.
  Offense+1.3380th
  Defense+1.1771st
• Rushing EPA

  EPA per designed run.
  Offense+0.2382nd
  Defense−0.1189th
• Passing EPA

  EPA per pass play.
  Offense+0.3574th
  Defense+0.1843rd
• Standard downs

  Success rate on downs where either a run or a pass makes sense, like 1st and 10 or 3rd and 2.
  Offense46.8%32nd
  Defense47.5%42nd
• Passing downs

  Success rate on downs that all but force a pass, like 2nd and 8 or longer, or 3rd and 5 or longer.
  Offense37.3%81st
  Defense20.5%96th
• Line yards

  Rushing yards per carry credited to the offensive line, weighting the first few yards most.
  Offense3.2265th
  Defense2.6381st
• Stuff rate

  Share of runs stopped at or behind the line of scrimmage.
  Offense17.1%55th
  Defense19.6%58th
• Havoc

  Share of snaps with a tackle for loss, sack, forced fumble, interception or pass breakup.
  Offense9.5%95th
  Defense16.8%62nd

Percentiles among the 138 FBS teams this season, where higher is better for the offense and the defense alike (a defense's numbers are what it allowed). The line in each bar is the FBS median. Garbage time excluded. Source: CollegeFootballData.

Drives vs. expectation, 2026

Every snap has an expected-points value: what a drive from that spot, down and distance is worth on average. A drive over expected beat its starting spot; defense counts what Duke allowed, so lower is better. The adjusted number fits every team against the opponents it actually played, so it leans toward average early in a season. Ranks are among FBS teams.

Game by game

1. W
   TULN
2. W
   ILL
3. W
   STAN
4. W
   W&M
5. L
   GT

Bars above the line: better than expected for Duke.

Players, 2026

1. Nate SheppardRB121 plays · +0.129 pts/play · 2nd of 320 RBs
   0.29WAR0.20 to 0.38
2. Walker EgetQB115 plays · +0.116 pts/play · 78th of 192 QBs
   0.25WAR0.06 to 0.41
3. Jeremiah HasleyTE28 plays · +0.191 pts/play · 1st of 164 TEs
   0.10WAR0.05 to 0.15
4. CJ Campbell Jr.RB33 plays · +0.102 pts/play · 159th of 320 RBs
   0.06WAR0.04 to 0.08
5. Kade ReynoldsonP13 plays · +0.076 pts/play · 35th of 84 punters
   0.02WAR0.00 to 0.04
6. Jake TaylorTE6 plays · +0.074 pts/play · 103rd of 164 TEs
   0.01WAR0.00 to 0.02
7. Nate KuriskyTE3 plays · +0.098 pts/play · 103rd of 164 TEs
   0.01WAR0.00 to 0.01
8. Javen NicholasWR1 plays · +0.170 pts/play · 341st of 364 WRs
   0.00WAR0.00 to 0.00

By position group, against FBS

Linemen and defenders are valued as units, in points per game against the median FBS unit, because play-by-play doesn't say who was on the field. Players are valued against a replacement-level backup, so the two aren't added together.

Wins above replacement credits each play's expected points to the players involved, adjusted for the opponent, and compares Duke's players with a typical backup; 53 points make a win. Only snaps while the game was in doubt count, so a starter on a team that blows opponents out plays fewer of them: read value per play beside WAR. The line under each player (or the range on a phone) is the 90% interval. Ranks are among FBS teams and players. How this works

What it had and what it did, 2026

• Roster talent

  247Sports' talent composite: the recruiting ratings of every player on the roster, transfers included.
  66866869th of 138Roster talent: 668, 69th of 138 in FBS
• Recruiting

  Average 247Sports class score over this season's class and the three before it.
  19619655th of 138Recruiting: 196, 55th of 138 in FBS
• Returning production

  Share of last season's production (expected points added) by players still on the roster.
  17%17%103rd of 136Returning production: 17%, 103rd of 136 in FBS
• SP+ rating so far

  The result: Bill Connelly's opponent-adjusted rating, points per game better than an average FBS team.
  +8.2+8.245th of 138SP+ rating so far: +8.2, 45th of 138 in FBS

Ranks and bars among the 138 FBS teams with a talent rating this season, best first; the line in each bar is the FBS median.

Talent conversion so far

Conversion by season

Fourth-down calls vs. the numbers, 2026

Each real fourth down is valued three ways (go for it, kick, punt) from the score, clock and field position before the snap, using league-wide odds rather than Duke's own. A call that differs gives up the gap to the best option; gaps under one point are close calls, and the 0 clear-cut ones account for 0.00 of the 0.02 expected wins. The model sees the situation, not the injury report, so a call it disagrees with can still be right. Ranks are among FBS teams; 1st gave up the least.

Costliest calls against the numbers

1. vs TULNQ3 9:47 4th & 2 at DUKE 20, up 10Punted; the numbers favored going for it (close call)−0.9 pts
2. at ILLQ1 1:36 4th & 2 at DUKE 20, down 3Punted; the numbers favored going for it (close call)−0.4 pts
3. vs TULNQ2 9:21 4th & 3 at TULN 14, up 7Kicked a field goal; the numbers favored going for it (close call)−0.3 pts
4. vs TULNQ2 0:00 4th & 5 at TULN 5, up 7Kicked a field goal; the numbers favored going for it (close call)−0.2 pts
5. vs TULNQ4 1:11 4th & 12 at TULN 38, up 14Went for it; the numbers favored a field goal (close call)−0.1 pts
6. vs W&MQ4 1:51 4th & 3 at DUKE 37, up 55Punted; the numbers favored going for it (close call)<0.1 pts
7. vs W&MQ1 7:45 4th & 5 at DUKE 26, up 7Went for it; the numbers favored punting (close call)<0.1 pts
8. vs STANQ4 1:52 4th & 3 at DUKE 33, up 28Punted; the numbers favored going for it (close call)<0.1 pts

Record vs. expectation, 2026

36th of 138 FBS teams at beating the model's pregame expectations

Expected wins adds up Duke's chance of winning each game as the model saw it at kickoff - a 70% favorite counts 0.7.

1. W

   TULN
2. W

   ILL
3. W

   STAN
4. W

   W&M
5. L

   GT

What moved it, 2026

Every game starts at Duke's pregame chance; everything after that is the field. The model's own breakdown of each game's final play splits the on-field part by what drove it, and the parts add up exactly. How it works →

• Scoreboard+0.74
• Possession & field position+0.02
• Clock+0.03
• Other−0.01
• Less the pregame edge, counted before kickoff−0.16

Game by game

• TULNW 17-3 vs TULN73% → 99%Before kickoff +23 · On the field +26: Scoreboard +28, Timeouts +1
• ILLW 31-27 at ILL31% → 96%Before kickoff −19 · On the field +65: Scoreboard +56, Clock +3
• STANW 35-7 vs STAN73% → 100%Before kickoff +23 · On the field +27: Scoreboard +28, Timeouts +1
• W&MW 62-7 vs W&M99% → 100%Before kickoff +49 · On the field +1: Scoreboard +1
• GTL 25-28 at GT62% → 5%Before kickoff +12 · On the field −57: Scoreboard −40, Timeouts −2

Points of win probability, from Duke's side.

Closest call

DUKE 46-45 CLEM (2025) - Duke dropped to a 2.6% win probability before pulling it out.