NFL + College FBS

Football

NFL + college FBS spreads and totals from a matchup-exploitation model: every unit is percentiled within its league, strong units are matched against weak opposing units (with an offensive-line layer and game-time weather), and the projection is anchored 65% to the de-vigged market. Picks are paper-only until they show positive closing-line value vs the sharp close. Records are always broken out by league and market — never blended.

Paper / simulatedPerformance →Live O/U →
Pick-mix alert: NFL totals lean 36% over across the last 50; CFB totals lean 66% over across the last 50 — outside the 60/40 band. Over/under bias check in order before trusting totals results.

Paper records · by league and market

NFL spreadsWins–losses–pushes on committed NFL spread paper picks (model matchup_v1). Exactly on the number = push, stake returned. Flat ROI -16.5% · avg CLV +0.28pp on 33 tracked.12-16-1
NFL totalsWins–losses–pushes on committed NFL total paper picks (model matchup_v1). Exactly on the number = push, stake returned. Flat ROI -12.8% · avg CLV +0.29pp on 59 tracked.23-30-0
CFB spreadsWins–losses–pushes on committed CFB spread paper picks (model matchup_v1). Exactly on the number = push, stake returned. Flat ROI +8.5% · avg CLV +0.40pp on 131 tracked.61-47-2
CFB totalsWins–losses–pushes on committed CFB total paper picks (model matchup_v1). Exactly on the number = push, stake returned. Flat ROI -13.5% · avg CLV +0.32pp on 193 tracked.78-93-0

Football board

Tue, Oct 20, 2026

AllNFLCollege

7:30 PM ET· 1 pick

South Alabama @ Marshall · CFB · week 8 · 7:30 PM ET

Over 65.5 (total) -102 (1.98)

PaperSimulated pick. Football is paper-only until it shows positive CLV vs the sharp close over a meaningful sample — never sized with real money.Dual edgeOne team holds BOTH the passing and rushing matchup edge — the model's highest-conviction archetype.moderate
EVRaw expected value from the blended cover/total probability vs the price. The threshold and tiers read ADJUSTED EV (sharp support/fade and fragile-edge haircuts applied once).+6.6%
Adj EVAdjusted EV: raw EV plus small signed context adjustments — sharp support/fade vs the sharp consensus and a fragile-edge haircut. Applied exactly once, here.+6.6%
ModelThe model's probability for THIS side, after blending 35/65 toward the de-vigged market — the same market-heavy anchoring philosophy as the MLB models.53.8%
MarketThe market's fair probability for this side — the price with the vig removed.48.1%
ConfidenceConfidence (0–100) from edge size and data completeness, with college-noise haircuts: CFB scores lower than NFL for the same edge, Group-of-Five games lower still, and explosive offenses cut totals confidence (variance, not projection).34
CLVClosing-line value vs the SHARP close, in probability points. + = the market moved toward our side after commit — the gate for football ever going live.+1.86pp
Projected South Alabama 38 — Marshall 33 · margin -4.4 · total 70.6 · stake 0.8u
Why this pick · matchup breakdown

Net edge -23.4 (Dual edge) — Marshall holds both the pass and rush matchup, the highest-conviction spot.

  • Marshall pass offense (32nd pct) vs South Alabama defense (21st pct): edge +11
  • Marshall rush offense (27th pct) vs South Alabama defense (13th pct): edge +14
  • South Alabama pass offense (54th pct) vs Marshall defense (27th pct): edge +27
  • South Alabama rush offense (65th pct) vs Marshall defense (17th pct): edge +48
  • Marshall OL (-6.3 pts): run-block proxy 4.43 ypc
  • South Alabama OL (+4.0 pts): run-block proxy 4.79 ypc
  • 62F, wind 5mph at kickoff
  • Line movement: sharp consensus 1.9pp toward over vs draftkings (line 65.5 vs 65.5)
  • Adjustment: noted: sharp consensus 1.9pp toward over vs draftkings (line 65.5 vs 65.5)
  • Fragile signal: OL layer on sack/YPC proxy only
Market: draftkings 65.5 (-102/-118) · sharp consensus 65.5 across 3 books · market mean 65.5 (model off by 5.1) · blend 35/65 model/market

5 other markets priced below the threshold and not shown.