Old Dominion at Virginia Tech. Lean: Old Dominion +19.5. Lean: Under 49.5.

Spread.Take Old Dominion +19.5, but discard the supposed 18-point model edge. EL's Virginia Tech -1.3 rests on questionable roster continuity assumptions and demonstrably incorrect coaching flags. Franklin's arrival and ODU's offensive departures justify much of the market disagreement. Still, the captured DraftKings move from -14 to -19.5 asks bettors to pay substantially more after one FCS game. Tech's 73-point opener establishes upside, not a reliable scoring baseline against FBS opposition. Henicle's running ability gives ODU a route to sustained possessions and enough scoring to stay within three touchdowns. The strongest counterargument is ODU's struggle to separate from Norfolk State, combined with its rebuilt receiving corps. I cannot establish a precise replacement fair spread from these samples, and no verified injury surprise creates an additional edge. This is a modest preference for the inflated underdog cushion, not confidence that the teams are close. Public preference for Tech does not independently establish value.

Total.Lean under 49.5 despite EL's 52.17 projection. ODU's quarterback transition and depleted receiving continuity are concrete reasons to discount the historical offensive baseline. Its opener remained competitive late, suggesting that 31 points overstated sustained dominance. The over case is substantial: Grunkemeyer was efficient, Tech demonstrated explosive scoring potential, and Henicle can create touchdowns with his legs. Neither the mild kickoff temperature nor unverified wind supports a major weather deduction. The captured decline from 52.5 to 49.5 has already consumed much of the under value, so conviction stays modest. I favor ODU struggling to contribute consistently and a potential Tech lead encouraging clock consumption; that game script is an inference, not verified tempo evidence.

Where our number looks wrong.The coaching flags are reversed: Virginia Tech has first-year coach James Franklin; ODU retains Ricky Rahne. EL also appears too anchored to Virginia Tech's poor 2025 team and ODU's successful 2025 offense despite substantial personnel changes. ODU's departed quarterback and top three receivers make near-parity particularly difficult to defend. Neither an FCS blowout nor the model's large gap establishes an ATS edge.

Spread.Our number (VT -1.3) is unusable. It is anchored to Old Dominion's 2025 SP+ off a 10-3 season and cannot see that the Monarchs return 9.2% of production, lost QB Colton Joseph to Nebraska, replaced five offensive line starters and their top three receivers. The market has the direction right; the 17.9-point away lean is an artifact, and I am not taking the dog because our model says so. I am taking it on price. DraftKings opened -14 (captured 9/2) and sits -19.5, with 72% of bets and 76% of money on Virginia Tech, so this is the top of a 5.5-point move. Preseason SP+ (VT 35th, ODU 95th) and FPI's 89.6% both price closer to 16-18 than 19.5. Virginia Tech is thin where a big number has to be earned: a first-year starter at quarterback, an offensive line Franklin says needs work, three receivers who missed week one, and thunderstorms arriving around halftime that argue for a run-out-the-clock second half. Reserves and a wet ball make 19.5 a live backdoor number.

Total.The dossier's captured forecast (74F, 3.5 mph, 30% precip) is stale. NWS on 9/9 has showers and thunderstorms Saturday, 80% for the day, a quarter to half inch of rain, 31% at the noon kickoff and 83% from 2pm - a wet, possibly delayed second half. Wind is near calm, so this is ball-handling and game script, not kicking. Both scripts shorten the game: Henicle threw 12 passes in week one and ODU wants to run, and Virginia Tech up three scores in the rain will milk clock and empty the bench. Beat projections average about 48 points. The counter is real: the total already fell from 52.5 and 74% of money is on the under, so most of this is priced. A lean, not a full-size position.

Where our number looks wrong.Badly wrong, and in a way worth fixing. The 17.9-point 'lean away, strong' is almost entirely the preseason prior reading ODU's 2025 SP+ (+5.9 off 10-3) against Virginia Tech's (-10.1 off 3-9), on a game where both rosters were rebuilt from the studs. ODU returns 9.2% of production with portal net -10 and the prior still rates them above a Virginia Tech team that hired James Franklin and imported Penn State's quarterback; the coaching-change term fires on ODU (flagged first-year head coach, which is factually wrong - Rahne is in year seven) and not on VT, where it belongs. The in-season refit then credits a 73-3 win over an FCS opponent with only 2.2 points of movement. Any gap this large against a 19-point market should be auto-suppressed when either side's returning production is under about 30%, or when the flagged coaching change is contradicted by the team page. I am on the same side as our model here by coincidence, not by agreement.

Spread.Take Old Dominion +19.5. But VT's 73-3 week-1 output came against FCS bottom-feeder VMI and says little about performance versus a real Sun Belt roster; ESPN FPI's [figure removed] does not obviously map to a 19.5-point number, and ODU has covered 57.1% of recent meetings with VT. Quinn Henicle remains a live dual-threat arm even with modest raw passing efficiency, and Rahne, a Franklin protege, knows this staff's tendencies. With several points of steam already bought by the market, I'd rather buy the hook than chase an already-shortened number.

Total.Lean under 49.5. ODU's Henicle attempted just 12 passes in the week-1 win, suggesting a run-first, clock-control approach that limits possessions, and a rebuilt ODU skill corps (gone: starting QB Colton Joseph, RB Trequan Jones) likely means fewer explosive plays than the 2025 offense produced. VT's 73-point output came against an overwhelmed FCS defense and won't repeat at that rate against a real Sun Belt defense that held Norfolk State to 10 points. Weather (74F, light wind, 30% precip chance per dossier capture) is not a material factor either way, but a large blowout with running clock also caps total plays.

Where our number looks wrong.Our -1.3 (VT) number is essentially a coin flip built on preseason priors, 2025 record (VT 3-9, ODU 10-3), returning production and blue-chip composite, and it under-weights two things the market has clearly priced: Virginia Tech's coaching upgrade (Franklin arriving after Pry was fired specifically because of last year's loss to this opponent) and Old Dominion's real offensive downgrade (lost its starting QB and leading rusher to the portal). The model has no mechanism for a revenge/program-reset narrative and no way to discount VT's 73-3 week-1 margin, which came against one of the weaker FCS opponents on the schedule. Given the model's own backtest shows no ATS edge at any gap size, a 17.9-point disagreement with the market should be read as the model missing real information, not as an exploitable inefficiency, which is exactly why I'm not fading the market by anywhere near the full gap.