Finding +EV bets is half the problem. Sizing them correctly is the other half. Bankroll management is the difference between compounding an edge and blowing up before it materializes.
A bankroll is the total amount of money you have set aside exclusively for betting. It is not your rent money. It is not your savings. It is a dedicated pool of capital that you can afford to lose entirely without it affecting your life.
This distinction matters because rational bet sizing requires emotional detachment. If you are betting with money you need for groceries, you will make emotional decisions: chasing losses, oversizing when desperate, or cashing out early when ahead. A separate bankroll eliminates those pressures.
The size of your bankroll determines everything else. It sets your unit size, your maximum exposure, your tolerance for drawdowns, and ultimately how long you can survive while waiting for your expected value edge to manifest. Even a profitable strategy will have losing weeks and losing months. Your bankroll needs to survive those stretches.
Professional bettors treat their bankroll like business capital. It has rules. It has limits. And the first rule is that preservation comes before growth. You cannot compound an edge if you go bust in week 6.
A “unit” is a standardized bet size, typically 1-2% of your total bankroll. If your bankroll is $5,000, one unit is $50-$100. Units let you measure performance independently of bankroll size. “I am up 12 units this season” means the same thing whether your bankroll is $1,000 or $100,000.
There are two primary approaches to unit sizing:
For most bettors, flat betting with tiered adjustments (see section 05) is the best balance of simplicity and edge optimization. Percentage-based sizing is theoretically optimal but introduces execution complexity that causes more problems than it solves for recreational bettors.
The Kelly Criterion is a formula from information theory that tells you the mathematically optimal percentage of your bankroll to wager on a given bet. It maximizes long-term growth rate while minimizing ruin probability.
Example: your model says a team wins 60% of the time (p = 0.60). The odds are +110 (b = 1.10).
Full Kelly says to bet 23.6% of your bankroll on this single bet. That is enormous — and that is the problem with full Kelly in practice. It assumes your probability estimates are perfectly accurate. They never are.
Half-Kelly is the standard adjustment: divide the Kelly recommendation by two. In this example, you would bet 11.8% instead of 23.6%. Half-Kelly sacrifices about 25% of the growth rate but reduces variance dramatically. It also provides a margin of safety for probability estimation errors. If your 60% is actually 55%, half-Kelly keeps you solvent. Full Kelly might not.
Most professional bettors use fractional Kelly (quarter to half). The exact fraction depends on how much confidence you have in your probability estimates. Models with strong calibration (predicted 65% wins at 65%) can use larger Kelly fractions than those with poor calibration.
A system that wins 58% against the spread at -110 odds is solidly profitable long-term. It will also, with near certainty, experience losing streaks of 5+ bets during any given NFL season. This is not a flaw. It is a mathematical guarantee.
Drawdowns — periods where your bankroll drops from its peak — are the price of playing a probabilistic game. A 58% bettor placing 80 bets per season will typically experience a maximum drawdown of 8-15 units at some point during the year. If your bankroll is only 20 units deep, a normal drawdown can wipe out 40-75% of your capital.
This is why bankroll sizing matters as much as pick quality. A 100-unit bankroll gives you the breathing room to survive the inevitable cold streaks. A 20-unit bankroll turns normal variance into a crisis that tempts you to abandon your process or chase losses.
Check the NoPunt results page to see the actual profit curve across multiple seasons. Every winning system’s equity curve has dips. The question is whether you are capitalized to ride through them.
NoPunt’s tier system grades every pick from S (highest conviction) to C (lean only). The tiers measure the gap between the model’s probability and the market’s implied probability on the spread or moneyline — a direct proxy for expected value. Larger gaps mean more edge, which justifies larger bet sizes.
Here is a suggested unit sizing framework that maps tiers to bet size:
| TIER | EDGE | UNITS | $50 UNIT |
|---|---|---|---|
| S | 8+ point implied edge | 2.0u | $100 |
| A+ | 5-8 point edge | 1.5u | $75 |
| A | 3-5 point edge | 1.0u | $50 |
| B | 1-3 point edge | 0.5u | $25 |
| C | Lean only | 0.25u | $12.50 |
This framework follows the Kelly principle: bet more when the edge is larger. S-tier picks have the widest model-vs-market gap and historically the highest hit rate, so they get the largest allocation. C-tier picks are marginal leans with minimal edge, so the bet size is minimal.
The “$50 unit” column assumes a $5,000 bankroll (1% = $50 per unit). Scale up or down based on your actual bankroll. The ratios between tiers matter more than the dollar amounts.
The hardest part of bankroll management is not the math. It is the psychology. Every impulse that makes sports fun — loyalty, excitement, the thrill of a big win — works directly against disciplined bankroll management.
A system removes emotion from the equation. When you have predefined rules for how much to bet on each tier, you do not need to make a decision in the moment. The model outputs a pick and a tier. The tier maps to a unit size. You place the bet. No deliberation, no second-guessing, no “I feel really good about this one so I am going 5x.”
This is why NoPunt publishes tiers, not just picks. The tier is not decoration. It is the sizing signal. An S-tier pick and a C-tier pick are both picks, but they are not the same bet. Treating them the same — or worse, sizing based on how excited you are about the matchup -- undermines the entire framework.
Here is a practical starting framework for NFL betting bankroll management:
The goal is survival first, compounding second. A bankroll that survives the variance is a bankroll that eventually shows the edge. A bankroll that blows up in week 8 never gets the chance.
NoPunt’s tier system tells you how much edge each pick has. Use it to size, not just to pick.