Bankroll Management: Flat, Kelly and How Much to Bet
Bet sizing decides whether your edge survives variance. Compare flat, percentage and Kelly staking, learn why pros use fractional Kelly, and pick a stake that won't blow up your bankroll.
You can find every good bet in the world and still go broke if you stake too much on each one. Bankroll management is the unglamorous part of betting that decides whether you last long enough for your edge to actually show up. Short-run variance is brutal. The whole point of sizing your bets well is to make a cold streak dent you instead of ending you.
Bankroll and the unit
Your bankroll is the money you've set aside for betting and nothing else. Not your rent, not next month's rent. Just a pot you're genuinely willing to swing up and down.
From that pot you define a unit, usually 1% to 3% of the bankroll. If your bankroll is 1,000 and a unit is 1%, then one unit is 10. Talking in units instead of raw cash keeps you honest. A "5-unit bet" means the same thing whether your bankroll is 500 or 50,000, and thinking that way stops you from suddenly betting triple your normal size to chase a loss.
Flat staking vs percentage of bankroll
There are two plain ways to size bets before you get into anything clever.
Flat staking means every bet is the same fixed amount, say 10 per bet, no matter the odds or how confident you feel. It's easy to track, and it stops you from talking yourself into a monster bet on a hunch. The downside is that it ignores your edge and it doesn't grow with your bankroll. Double your money and you're still betting the same 10.
Percentage-of-bankroll staking means you bet a fixed slice of your current bankroll, say 1%, so the stake rises when you're winning and shrinks when you're losing. That built-in shrinkage is protective. You never end up betting a fixed cash amount into a pot that's half the size it used to be.
| Approach | Stake on 1,000 bankroll | After a run down to 600 |
|---|---|---|
| Flat (10 per bet) | 10 | 10 |
| Percentage (1%) | 10 | 6 |
Flat is simpler. Percentage adapts. Most people who bet seriously drift toward some percentage-based method, because it responds to reality instead of ignoring it.
The Kelly criterion
Kelly is a formula for the stake that maximizes long-run growth given your edge. It needs one hard input: an honest estimate of your true win probability, p. That's the catch, and it's a big one. If your p is off, Kelly will happily tell you to bet way too much.
For decimal odds O, the math is short:
b = O - 1(your net profit per unit staked if the bet wins)edge = p * O - 1(your expected value per unit)- full Kelly fraction
f = edge / b
Say you rate a selection at p = 0.55 and the odds are 2.00. Then b = 1.00, and edge = 0.55 * 2.00 - 1 = 0.10. Full Kelly is 0.10 / 1 = 0.10, which means 10% of your bankroll on that single bet.
Ten percent on one bet. If that makes you a little nervous, good. It should.
Why serious bettors use fractional Kelly
Full Kelly maximizes growth on paper, but it assumes your probability is exactly right and it swings hard. A short run of bad luck at 10% a pop can halve your bankroll fast. Since almost nobody's p is perfect, full Kelly tends to overbet reality.
The fix most sharp bettors reach for is fractional Kelly: take a fraction of the full-Kelly stake, usually somewhere around 20% to 30%. Quarter-Kelly of the example above is 0.10 * 0.25 = 0.025, so 2.5% of bankroll instead of 10%. You give up a small slice of theoretical growth and get a large drop in variance in return. For most bettors that's an easy trade to make.
Risk of ruin
Being +EV doesn't protect you from going broke. Risk of ruin is the chance a losing streak wipes you out before your edge ever pays off, and it climbs steeply with stake size. Bet 2% a time and a rough patch is survivable. Bet 20% a time and a perfectly normal, expected losing run can end your bankroll while your edge was real the whole way through.
So the quiet danger of over-staking is this: the bets can be good and the bettor can still bust. Small stakes buy you the one thing an edge needs, which is time.
Scaling with time to kick-off
Here's one extra wrinkle. Not every bet deserves the same fraction. A price you take a week before kick-off has far more room to move against you than one you take minutes before the market closes, when the odds are sharpest. A curated approach can lean into that by staking a smaller fraction on early bets and a larger fraction as the event approaches and your read sits on firmer ground. It rewards patience without asking you to run the numbers by hand every time.
All of this only works alongside the rest of the process. If you're not sure a bet is +EV in the first place, start with value betting and positive EV. To check whether you're actually beating the market, read up on closing line value.
Betamigos turns all of this into one number per bet. From your bankroll, your edge and a safety cap you set, it suggests a stake using flat, Kelly or a curated method that scales with time to kick-off, so you're not eyeballing percentages in your head. See it in your dashboard.
Betamigos is an odds-analytics tool, not a sportsbook, and none of this is a promise of profit. Bet responsibly. 18+.