Five things bettors “know” about football. Guess the real rate, then see 2023-2025 of actual games. Most edges land near a coinflip — the market already priced them. The gap between your guess and reality is the leak.
Night games are sloppy/defensive — bet the under.
If unders had an edge, every game would show it.
Familiarity + defense — division unders.
Perfect conditions — bet the over indoors.
Good teams at home blow the doors off — lay the points.
Four of these five are coinflips — the “edge” is already in the number. The one with a lean (primetime unders) is still inside the noise for a 3-season sample. This is the whole game: your read has to beat the price, not just sound true. We grade against the closing line for exactly this reason.
All three arms lose on net — favorites are efficiently priced, so there’s no alpha. But the sportsbook blows up 43.8% of weekends and the unmanaged prediction “crossover” 18.8%; disciplined management takes ruin to 0% at the best net of the three.
¹ Blow-up frequencies assume equal flat sizing across all three arms; the sizing-independent result is the relative compression — drawdown ~4×, variance ~13×, ruin → 0%, robust across all 16 Sundays.
² Figures are real, 16 qualifying 2025 NFL Sundays, no rounding to any hypothesis. 1-min granularity. Analytics & education — not advice, not picks.
Source: nflverse schedules · 2023-2025 · 855 games. Research only — not betting advice.