Bankroll & staking, done properly

Here's the uncomfortable truth of betting maths: you can have a genuine edge and still go broke. Bet too large a slice of your money each time and a perfectly normal losing run — the kind arithmetic guarantees will happen — wipes you out before the long run arrives to pay you. The reverse is also true: modest edges, staked correctly, compound into serious returns because you're always in the game when the maths pays out. Every professional bettor will tell you the same thing — selection gets the glory, staking pays the bills. This guide gives you the complete system: how big the bankroll, how big the bets, and the exact reasoning, with numbers you can copy today.

Your bankroll is a fixed sum, decided once, in cold blood: money that could go to zero without touching your rent, bills, or peace of mind. £100 and £5,000 run the identical system — percentages don't care. Sealed box means sealed both ways. Nothing goes in impulsively: topping up mid-losing-run ('reloading') is the single most reliable symptom of chasing, and chasing is how betting stops being maths and starts being a problem. Nothing is 'borrowed' out either — the bankroll's size determines every stake, so raiding it quietly breaks the whole system. Schedule deposits and withdrawals like a business — monthly, decided in advance, when you're calm. And if the sealed box ever stops feeling like entertainment money, that's the signal to visit our responsible gambling page, not the deposit page.

A unit is 1% of your bankroll. £500 bankroll → £5 unit. From now on, every stake is expressed in units: standard bets are one unit, your strongest positions maybe two or three, and nothing — ever — exceeds five. What this buys you: proportionality (as the bankroll grows or shrinks, stakes scale automatically — you compound wins and defend automatically in downswings), comparability (your log becomes readable: '+11 units over 300 bets' is a real record; '£77 up' means nothing without context), and calm (a bet that's 'one unit' triggers none of the emotional weight of the same bet framed as dinner money). Recalculate the unit monthly, or after any ±25% bankroll move. Takes one minute.

At value-betting odds you'll win perhaps 40% of bets. Here's what streaks look like at that win rate — not worst case, just normal arithmetic over a few hundred bets: • 5 straight losses: routine — expect it monthly. • 8 straight: happens to everyone playing long enough — almost certain inside a year of regular betting. • 10–12 straight: genuinely will happen across a betting career, to bettors doing everything right. Now the punchline. At 5% of bankroll per bet, a 12-loss streak costs about 46% of your money — most people's discipline dies there and the spiral finishes the job. At 1% per bet, the same streak costs about 11% — annoying, survivable, forgettable. Same edge, same bad fortnight, completely different outcomes. That's the entire argument for small stakes, in one table.

Flat one-unit staking is already good. Kelly is the refinement: stake proportional to how big the edge actually is. The formula: stake (as a fraction of bankroll) = edge ÷ (odds − 1). Worked example: the scanner flags a bet at odds 2.50 with true probability 44%. Edge = (0.44 × 2.50) − 1 = +10%. Kelly = 0.10 ÷ 1.50 = 6.7% of bankroll. Smaller edge, smaller stake: a +3% edge at the same odds gives 2% of bankroll. Longer odds shrink stakes too, automatically — a +10% edge at odds 6.00 gives just 2%, because long shots need smaller positions for the same growth. Kelly is mathematically the fastest-compounding staking scheme that never risks ruin. But it has one dangerous assumption, and the next step deals with it.

Full Kelly assumes your edge number is exactly right. In real betting it never is — 'true probability' is an estimate, and when your estimate runs slightly hot, full Kelly doesn't just underperform, it systematically over-stakes, and over-staking an overestimated edge is a bankroll shredder. Full Kelly also swings violently even when the edge IS right: 50% drawdowns are within normal experience. The professional consensus: bet a fraction — a quarter to a half of the Kelly number. Quarter Kelly keeps roughly three-quarters of the long-run growth rate while cutting drawdowns to a fraction. That trade is a bargain, which is why the scanner's suggested stakes come pre-capped at quarter-Kelly: the number on screen is one you can stake as shown. So the practical system, complete: 1% flat units while you're new → scanner's fractional-Kelly suggestions once the routine is automatic → nothing over 5% of bankroll, ever, no matter what.

1. Sealed bankroll — affordable-to-lose, no tilt top-ups, no quiet raids. Deposits/withdrawals scheduled, not emotional. 2. Units — 1% each, every stake in units, recalculated monthly. 3. Fractional Kelly — the scanner's capped suggestion or your own quarter-Kelly; 5% absolute ceiling on any single bet. 4. Never chase — after a loss, the next stake is whatever the system says, not a recovery bet. If you feel the pull, stop for the day; the fixtures will still be there tomorrow. 5. Log everything — every bet, every stake, every price. The log is what lets you trust the process through a normal bad month, and it's the honesty mechanism that keeps you within your edge. Boring, repeatable, and the exact difference between people who have an edge and people who keep one.