Your edge means nothing if you go broke before it pays off. Bankroll management is the discipline that keeps you in the game long enough for the math to work.
Here's a truth that surprises most bettors: you can have a genuine edge and still go broke. A bettor who wins 55% of their spread bets (a very strong record) can easily lose their entire bankroll if they bet too aggressively during an inevitable cold streak.
Variance is real and it's brutal. Even with a 55% win rate on -110 spreads, a 10-bet losing streak will happen roughly once every 2,700 bets. If you're betting 10% of your bankroll per game, that losing streak wipes out 65% of your account. Most bettors don't survive that psychologically, even if the math says they'd recover eventually.
Bankroll management solves this problem. It's the framework that determines how much of your money goes on each bet, ensuring you can absorb the inevitable downswings while still maximizing long-term growth.
Your bankroll is money set aside exclusively for betting. It is not your rent money, your savings, or your "I'll replace it next paycheck" fund. It's capital you can afford to lose entirely without affecting your life.
For most recreational bettors, this means starting with $500 to $2,000. For more serious bettors, it could be $5,000 to $25,000. The exact number matters less than the principle: this money is separate from everything else.
A "unit" is your standard bet size, expressed as a percentage of your bankroll. This is the foundation of every bankroll management system.
| Bettor Type | Unit Size | Example ($1,000 bankroll) |
|---|---|---|
| Conservative | 1% of bankroll | $10 per bet |
| Standard | 2-3% of bankroll | $20-$30 per bet |
| Aggressive | 5% of bankroll | $50 per bet |
| Reckless | 10%+ of bankroll | $100+ per bet |
Most professional sports bettors use 1-3% per bet. If that sounds small and boring, good. Boring is what keeps you solvent through a 15-game losing streak.
The math is clear on this. At 2% per bet, even a catastrophic 15-bet losing streak only costs you 26% of your bankroll (because each subsequent bet is 2% of a shrinking total). At 10% per bet, that same streak costs you 79%.
There are two main approaches to unit sizing, and each has merit depending on your situation.
If you're just getting started, flat betting is the right call. Pick a unit size (2% is a solid default) and bet the same amount on every play. As you build a track record and gain access to tools that estimate true probabilities, you can graduate to variable sizing.
The Kelly Criterion is the mathematical formula for optimal bet sizing. Given your estimated probability of winning and the odds being offered, it tells you exactly what percentage of your bankroll to wager for maximum long-term growth. Calculate optimal bet size with our Kelly Criterion Calculator by entering your bankroll, odds, and win probability estimate.
The formula is: Kelly % = (bp - q) / b
For example, if you estimate a bet has a 58% chance of winning at -110 odds (b = 0.909):
Kelly % = (0.909 × 0.58 - 0.42) / 0.909 = 11.6%
That's the full Kelly recommendation. In practice, nearly every professional bettor uses fractional Kelly (typically quarter or half Kelly) to reduce variance. So this bet would get sized at roughly 3-6% of your bankroll.
Even with proper unit sizing, you need hard stops for when things go wrong. Drawdown limits are predetermined points where you pause, reduce bet sizes, or stop entirely.
| Drawdown Level | Action |
|---|---|
| Down 10% from peak | Review your process. Are you following your strategy? Recalculate units based on new bankroll size. |
| Down 20% from peak | Reduce unit size by 50%. Focus only on your highest-confidence plays. |
| Down 30% from peak | Stop betting for at least one week. Review every bet placed during the drawdown. Identify leaks. |
| Down 50% from peak | Full stop. Something is fundamentally wrong with your approach or your market has shifted. Do not resume without a complete strategy review. |
These limits serve a dual purpose. They protect your capital, and they protect your psychology. The worst decisions in sports betting happen when someone is trying to dig out of a hole.
Modern AI research tools have made bankroll management significantly easier. Instead of guessing how confident you should be in a bet, tools like Juice provide probability estimates and expected value calculations for every bet you analyze. The core principle is simple: only bet when EV is positive. Every dollar committed to a negative expected value bet is a long-run loss regardless of whether that individual bet wins.
This matters because the Kelly Criterion requires a probability estimate as its input. If your probability estimate is bad, your bet sizing will be bad too, even if you're using the formula correctly. AI research tools dramatically improve the quality of that input.
Here's what a proper AI-assisted workflow looks like:
Juice uses AI to analyze any bet from a screenshot, giving you the probability estimates and EV calculations you need for proper bankroll management.
Try Juice Free for 3 Days →After a losing streak, the temptation is to increase bet sizes to "make it back faster." This is how bankrolls die. Your unit size should be based on your current bankroll, not on what your bankroll used to be.
If your bankroll drops from $1,000 to $800, your 2% unit drops from $20 to $16. Many bettors keep betting $20 because it "feels right." That's now a 2.5% unit, which compounds the problem during continued losses.
Parlays are fun and occasionally they hit big. But from a bankroll management perspective, they're terrible vehicles. The vig compounds across each leg, giving the sportsbook a massive edge. If you insist on playing parlays, limit them to a tiny fraction (0.5%) of your bankroll and treat them as entertainment, not strategy.
You can't manage what you don't measure. Every bet should be logged: date, sport, bet type, odds, stake, result, and your estimated probability at the time. After 200+ bets, patterns emerge that tell you where your edge actually is (and isn't).
Saying "I'm really confident in this one" and tripling your normal bet is not a strategy. Confidence should be expressed as a probability estimate and fed into a formula (like Kelly), not used as a vague justification for oversizing.
If all of this feels like a lot, here's the simplest version that still works:
That's it. This plan won't maximize your theoretical growth rate, but it will keep you alive through variance while you develop your skills and build a track record. Once you have 500+ tracked bets and a demonstrated edge, you can graduate to fractional Kelly sizing with AI-assisted probability estimates. For a broader list of tools to support your bankroll process, see our guide to the best free tools for sports bettors.