Bet sizing decides who survives an NFL season, more than picking winners does. The schedule hands you 17 games per team — a laughably small sample — which means variance batters even genuinely good judgements, and staking mistakes get exposed brutally. If the relationship between odds, vig and probability is still hazy, start with the NFL betting guide; this piece is about protecting money once you understand what you are buying.
How big should one NFL bet be?
The standard answer holds up: a unit of 1-2% of bankroll, staked flat, with your most confident bets earning at most 2-3 units rather than five or ten. The reasoning is arithmetic, not caution for its own sake. At -110 you need 52.4% winners just to break even, and the best documented NFL bettors sustain something like 54-56% over large samples. At a 55% true win rate, losing streaks of five and six are routine across a season, and a bettor staking 10% per bet gets wiped out by ordinary variance while a 1-2% bettor barely notices.
The NFL adds its own trap: scarcity. One main slate a week makes each Sunday feel enormous, and stake inflation follows emotional importance. The week after a bad Sunday is exactly when discipline pays — chasing losses with bigger stakes converts a bad week into a bad season. Flat staking exists to make that decision automatic instead of emotional.
Why do key numbers make NFL spreads unusual for value hunting?
Because NFL margins are not smooth. Roughly 15% of games land exactly on 3 and around 9% land exactly on 7, a consequence of field goals and touchdowns deciding close professional games. No other major sport concentrates outcomes this hard, which changes what a half point is worth: the half point between -2.5 and -3 covers the most common final margin in the sport, while the half point between -4.5 and -5 covers almost nothing.
The cheapest way to exploit this is line shopping. Different books post different numbers, and taking -2.5 at one book instead of -3 at another costs no vig at all — it is pure improvement. Over a season, systematically getting the best side of 3 and 7 is worth more than most people's handicapping. It also compounds with everything else: futures and win totals reward shopping the same way, and prop markets punish non-shoppers even harder given their wider margins — see the player props guide for those numbers.
When is buying points worth the price?
Books let you move a spread half a point in your favour for a fee, typically 10 cents of odds per half point — except around the numbers that matter, where the fee jumps because the books can also count.
| Move | Typical cost | Long-run verdict |
|---|---|---|
| -3 to -2.5 (crossing 3) | 20-25 cents | roughly fair at 20, marginal at 25 |
| +2.5 to +3 (onto 3) | 20-25 cents | same logic, same verdict |
| -7.5 to -7 (onto 7) | 10-15 cents | close to fair at 10 |
| -5 to -4.5 (dead number) | 10 cents | clear donation |
The logic: landing on 3 happens often enough that converting a loss to a push, or a push to a win, has measurable value — worth roughly 20-ish cents of odds. Books price the crossing at 20-25 cents precisely because that is what it is worth; you are trading at fair value, not stealing. Away from key numbers, the same 10-cent fee buys nearly nothing, because margins land on 4, 5 or 9 far less often. The rule of thumb: buying across 3 is defensible, buying across 7 is marginal, buying anywhere else is giving the book a tip.
How do you stay disciplined across a 17-game season?
The structural leaks are behavioural, and they are predictable:
- Forced Sunday action. Betting every week because the slate exists is the biggest leak in NFL betting. Some weeks offer nothing; passing is a position.
- Parlay drift. Each -110 leg added to a parlay compounds the margin — a two-leg parlay holds around 9%, and heavily promoted same-game parlays can hold 20-30%. Books market parlays for a reason.
- Untracked bets. Log every bet with its closing line. If you consistently beat the close, your process works even during losing runs; if you do not, no hot month changes the diagnosis.
- Stake creep during live betting, where adrenaline and wider vig gang up on discipline.
- Small-sample overreaction — 17 games per team means September records are mostly noise, and pricing follows the noise faster than you can.