The math they hide
You don’t need a conspiracy. You need a calculator. Every number below comes from 2855 NFL games with closing lines since 2016 ([src]), computed by us — open data, reproducible.
The 52.38% wall
Flip a coin against the book at −110 and you lose 4.5 cents of every dollar, forever. To just tread water you have to be right 52.38% of the time. Almost nobody is, over thousands of bets.
Average built-in margin at NFL close (2016–now)
| Market | Games | Avg hold |
|---|---|---|
| Point spread | 2855 | 3.13% |
| Moneyline | 2855 | 3.09% |
| Total | 2855 | 3.23% |
| Moneyline, fav Pick’em to −150 | 739 | 2.97% |
| Moneyline, fav −151 to −250 | 1076 | 3.07% |
| Moneyline, fav −251 to −400 | 629 | 3.10% |
| Moneyline, fav −401 or bigger | 411 | 3.33% |
That is the margin on the sharpest, most-bet market in America at closing. Props, alt lines and parlays are worse — see Parlays.
Favourites are a tax on fans
Closing favourites won straight up 66.3% of the time. Against the spread they went 1350-1431-71 (48.5%). The public loves favourites; the line makes them pay for it.
| Season | Fav ATS | Cover% |
|---|---|---|
| 2016 | 137-124 | 52.5% |
| 2017 | 133-123 | 52.0% |
| 2018 | 118-140 | 45.7% |
| 2019 | 120-137 | 46.7% |
| 2020 | 119-150 | 44.2% |
| 2021 | 131-150 | 46.6% |
| 2022 | 124-150 | 45.3% |
| 2023 | 141-130 | 52.0% |
| 2024 | 148-133 | 52.7% |
| 2025 | 138-146 | 48.6% |
| 2026 | 41-48 | 46.1% |
Long shots are where they bleed you hardest
Moneyline price vs what actually happened. Implied = the book’s price (vig included). The bigger the underdog, the worse the deal — the classic favourite–longshot bias, in the NFL’s own numbers.
| Price band | Bets | Book implied | Actually won | ROI betting it blind |
|---|---|---|---|---|
| Fav −400 or more | 420 | 85.5% | 84.0% | -1.8% |
| Fav −250 to −399 | 648 | 75.5% | 75.3% | -0.3% |
| Fav −150 to −249 | 1062 | 65.3% | 62.2% | -4.8% |
| Fav −101 to −149 | 805 | 55.6% | 53.7% | -3.6% |
| Dog +100 to +149 | 926 | 44.8% | 43.4% | -3.2% |
| Dog +150 to +249 | 1006 | 34.8% | 34.6% | -1.3% |
| Dog +250 to +399 | 574 | 24.7% | 22.3% | -9.3% |
| Dog +400 or more | 245 | 15.1% | 13.1% | -16.8% |
A +400 dog is priced like it wins 1 in 6.6. Since 2016, NFL dogs of +400 or more lost 16.8% of every dollar bet on them. That is the product the books push hardest in their ads.
What betting blind cost: $100 a game since 2016
| Strategy | Bets | Profit / loss | ROI |
|---|---|---|---|
| Every home team, spread | 2856 | -$13,550 | -4.7% |
| Every road team, spread | 2856 | -$3,893 | -1.4% |
| Every favourite, spread | 2856 | -$16,035 | -5.6% |
| Every underdog, spread | 2856 | -$1,389 | -0.5% |
| Every over | 2856 | -$14,725 | -5.2% |
| Every under | 2856 | -$3,506 | -1.2% |
| Every home team, moneyline | 2855 | -$14,503 | -5.1% |
| Every road team, moneyline | 2855 | -$8,383 | -2.9% |
| Every favourite, moneyline | 2855 | -$9,004 | -3.2% |
| Every underdog, moneyline | 2855 | -$13,881 | -4.9% |
| Both sides of every spread AND total | 11424 | -$35,674 | -3.1% |
Every single blind strategy loses. Bet both sides of everything and the book pockets +$35,674 for doing nothing but booking your action. That’s the business model.
Parlay math
Each −110 leg pays 1.909× instead of a fair 2.00×. Stack them and the house edge compounds:
| Legs | Book pays (decimal) | Fair payout (50/50 legs) | Built-in house edge |
|---|---|---|---|
| 1 | 1.91× | 2× | 4.5% |
| 2 | 3.64× | 4× | 8.9% |
| 3 | 6.96× | 8× | 13.0% |
| 4 | 13.28× | 16× | 17.0% |
| 5 | 25.36× | 32× | 20.8% |
| 6 | 48.41× | 64× | 24.4% |
| 8 | 176.45× | 256× | 31.1% |
| 10 | 643.08× | 1,024× | 37.2% |
| 13 | 4,474.51× | 8,192× | 45.4% |
A 10-leg −110 parlay hands the book a 37.2% edge before you even pick a side. Regulators’ own numbers below confirm it.