← Free tools

Bankroll calculator

Your unit size — and how long it survives a cold streak.

Your unit (1 bet)

20.00

Flat: losses to halve

25

Proportional: losses to halve

35

At 2% your unit is 20.00conservative — the standard range. Flat staking, it would take 25 straight losses to halve your bankroll (35 if you re-size to the smaller bankroll each time).

Unit size compared: stake and consecutive losses needed to halve the bankroll
UnitStake per betFlat: losses to halveProportional
0.5%5.00100139
1%10.005069
2%20.002535
3%30.001623
5%50.001014
10%100.0057

“Losses to halve” counts consecutive losing bets — an illustration of how much room a unit size leaves you, not a probability. It deliberately isn’t risk of ruin, which needs an estimate of your edge and your variance. Real losing streaks arrive mixed with wins, so the practical lesson is the ordering, not the exact count.

How to size a betting bankroll

Your bankroll is money set aside exclusively for betting — an amount you could lose entirely without touching your real finances. Your unit is a fixed slice of it, and every staking decision is expressed as a percentage of the bankroll rather than a dollar figure you feel like betting.

The consensus range is 1–2% per bet. That isn’t timidity — it’s arithmetic. At 2% it takes 25 consecutive losses to halve your bankroll; at 10% it takes five. Since even a genuinely profitable bettor loses 45% or more of their bets, and losing runs of six or eight arrive routinely, the unit size you pick decides whether variance is an inconvenience or the end of your bankroll.

The second decision is flat versus proportional. Flat bets the same amount every time — simple and immune to overconfidence. Proportional re-sizes to your current bankroll, so it automatically shrinks stakes when you’re losing and compounds them when you’re winning. Either beats the pattern that actually busts people: raising the stake after a loss to win it back.

Common questions

What percentage of your bankroll should you bet?
For most bettors, 1–2% of your bankroll per bet. That's small enough to survive a normal losing streak and large enough to matter. Bettors with a measured, proven edge sometimes go to 3–5%, but staking above 5% per bet risks losing half your bankroll to ordinary variance alone.
How do you calculate a betting unit?
Unit = bankroll × unit percentage. On a $1,000 bankroll, a 1% unit is $10 and a 2% unit is $20. Recording results in units rather than dollars keeps your record comparable across bankroll sizes — “+12 units” means the same thing to anyone.
How much money do you need to start a betting bankroll?
Whatever you could lose entirely without it affecting your finances — that's the only rule that matters. Practically, a bankroll needs to be large enough that a 1–2% unit is still a stake you can place, so many sportsbooks' minimums put that around $100–$500.
What's the difference between flat and proportional staking?
Flat staking bets the same dollar amount every time. Proportional staking bets a fixed percentage of your current bankroll, so stakes shrink during losing runs and grow during winning ones. Proportional survives a cold streak noticeably longer — at a 5% unit, 14 straight losses to halve versus 10 flat.
Is a bankroll calculator the same as the Kelly criterion?
No. A bankroll calculator sizes every bet the same way regardless of the specific bet. Kelly sizes each bet to the edge you believe it has, so it needs an honest win-probability estimate. Use this for a baseline staking plan, and Kelly when you're pricing a particular bet.

Next: size a specific bet to its edge with the Kelly criterion calculator, read bankroll management 101, or see why chasing losses is the costliest habit in betting. Glossary: unit · bankroll management · risk of ruin

If betting stops being fun or feels hard to control, help is available: ConnexOntario 1-866-531-2600 · ncpgambling.org.

Track every bet automatically

Stop doing the math by hand. Bankroll Guardian logs your bets and shows P&L, ROI, CLV and more — free to start, no card required.