William & Mary at Duke. No pick on the spread and no pick on the total.

Spread. The strongest explanation for Duke -43.5/-44 is substantial superiority on both lines: Duke allows 12.3 points per game, while William & Mary surrendered 96 points across its last two games. Those weaknesses are visible and plausibly priced into the [web market](https://www.covers.com/sport/football/ncaaf/matchup/378026/odds). Our model supplies no spread because William & Mary is unrated; there is no numerical disagreement to exploit. The strongest away case is possession compression and substitution risk. Duke's Stanford victory included four touchdown drives lasting more than four minutes, and this week's reporting discusses creating opportunities for younger players. Watkins also gives the Tribe a credible route to a late touchdown. However, those considerations do not establish that 44 points is excessive: Duke's rushing advantage could generate touchdowns without needing aggressive play calling. Gardner's stale injury entry does not establish a starting-quarterback absence; Eget heads the available depth chart. I have no actionable ATS view, and the empty captured receipt independently prevents panel support.

Total. The strongest explanation for 62.5 is Duke's potential to overwhelm a defense that just conceded 44 and 52, with Watkins providing enough offense to supplement the favorite. Favorable weather supports that over case. The under argument has concrete support: Duke allowed Stanford 237 yards and possessed the ball for 37:14, demonstrating a route to control without excessive possessions. Those [official game statistics](https://goduke.com/news/2026/9/19/football-duke-wins-fifth-consecutive-acc-opener-defeat-stanford-35-7) matter more than generic blowout assumptions. Our 57.74 projection is only 4.76 below 62.5, well inside its 12.86-point historical error. Action's displayed 61.5 versus a 62.5 opener suggests some under adjustment, but conflicting snapshots prevent a verified movement claim. Explosive runs, turnovers and reserve-defense scoring could defeat the under. Neither direction has sufficient incremental evidence, and no captured total exists.

Spread. There is no ATS quote in available_quotes, so this is a mandatory off-market pass regardless of view. For context, the market elsewhere sits at Duke -43.5 vs FCS William & Mary. The strongest explanation for that number is simple class gap: Duke is 3-0 with a stingy defense (12.3 PPG allowed, held Tulane to 3 and Stanford to 7), while W&M is a 2-2 FCS side whose defense has collapsed lately, surrendering 44 to Lehigh and 52 to Fordham (both FCS). That says Duke can win by a lot, but 43.5 is an enormous number where garbage-time clock control, a running-clock second half and pulled starters routinely shrink margins. Our own spread model produces no number here (W&M unrated) and has no proven ATS edge, so it adds nothing. I hold no genuine read strong enough to lay 43.5 or to buy back the points, and with no captured quote there is nothing to bet into. Pass, no view.

Total. No total quote in available_quotes, so this is a mandatory off-market pass. Context: the market elsewhere is around 62.5. Our totals model predicts 57.74 (with a wind drag of about 1.85 points at 9.4 mph), sitting under the market by roughly 4.8 points, well inside its own 12.86-point noise threshold, so it names no side. Weather is benign (clear, 76F, low wind). The directional case for an under is that Duke's defense is genuinely good (12.3 PPG allowed) and a huge favorite often plays a slower, clock-milking second half with backups, which caps scoring even in a blowout. The case for the over is W&M's leaky defense and up-tempo shootout profile (52-27, 44-20 recent games). These roughly offset, the model gap is noise-sized, and there is no captured number to bet. Pass, no view.

Spread. The market's -43.5 for Duke looks driven by a stark schedule-quality gap: Duke is 3-0 with a +15.4 per-game scoring margin against Tulane, Illinois and Stanford, while William & Mary is 2-2 in the Patriot League with a -9.8 margin, including blowout losses to Lehigh (20-44) and Fordham (27-52) that expose a defense allowing 35.8 ppg. Duke's QB1, grad transfer Walker Eget, is healthy and has started all three games (539 yds, 4 TD, 1 INT), so there's no QB-change story here; a Covers entry flagging 'L. Gardner' as an injured Duke QB appears stale, tied to language about last week's already-played Stanford game, not this one, and no source names Gardner as a current Duke quarterback. Nothing found this week moves the number off what the underlying scoring gap already supports. No ATS quote exists in this run's available_quotes, so this is an off-market pass regardless of lean; the counter-case (backup rotation, garbage-time W&M scoring) keeps any directional read from clearing a bettable bar anyway.

Total. No total quote exists in available_quotes, so this is a pass regardless of read. Directionally it's close: Duke's offense (27.7 ppg vs real competition) against a William & Mary defense that just allowed 44 and 52 points to Lehigh and Fordham could push Duke's own output well past 62.5 alone, arguing over; but a 43.5-point market spread usually implies backups playing much of the second half, which tends to cap live scoring below what full-strength averages suggest, arguing under. I found no confirmed word on Duke's rotation plans, and no weather issue (forecast is mild, clear, light wind) that would tip this either way, so I have no real edge over the market's own total.