Bankroll management and the case for quarter Kelly
By Adam · Updated 16 July 2026 · 6 min read
You can find genuine value and still go broke. It's the least advertised fact in betting: an edge describes the long run, and staking decides whether you're still solvent when the long run arrives. This guide covers the honest options, from flat stakes to the Kelly criterion, and why this site surfaces quarter Kelly rather than the full version.
What a staking plan is actually for
Two jobs. The visible one is growth: staking more when the edge is bigger, so good judgement compounds. The invisible one matters more: a staking plan is a decision you make in advance, while you're calm, about what you'll do during a losing run. Every bettor eventually hits the run that makes doubling up feel sensible. The plan's real value is that you already answered that question before the run started.
Flat percentage stakes, the honest baseline
Stake a fixed 1–2% of your current bankroll on everything you rate as value. No formula, no tuning, nothing to get wrong at midnight. Because the stake is a percentage, it shrinks in losing runs and grows in winning ones automatically, which handles most of what fancier schemes promise. Plenty of people who take betting seriously should stop reading here and use this; it captures most of the benefit of everything below, with far fewer ways to hurt yourself.
Kelly, for the curious
The Kelly criterion answers a precise question: what fraction of a bankroll maximises long-run growth if my probability is exactly right? The formula is the edge divided by the odds minus one. Take a leg you rate 40% at odds of 2.75: the edge is 10%, the odds-minus-one is 1.75, and Kelly says stake 5.7% of the roll. Bigger edges and shorter odds push the stake up; thin edges at long prices push it down to almost nothing. As a way of ranking how much money a bet deserves, the logic is hard to fault.
As an instruction, it has a flaw you can now spot on sight: if my probability is exactly right. Yours isn't. Every input is an estimate, and Kelly takes estimates literally.
Why full Kelly bites
Overestimate your edge by half — rate a leg 40% when the truth is 38% — and full Kelly doesn't merely grow slower. It can stake a winning process into a losing one, because oversized stakes drag on a bankroll faster than the real edge replenishes it. Model edges are precisely the kind that get overestimated: data noise, a market that got sharper since the last fit, the gap between a backtest and live prices. Full Kelly also swings violently even when the inputs are honest, with drawdowns deep enough to make a disciplined person abandon a working strategy at the exact bottom.
Quarter Kelly is the standard repair, and the trade is asymmetric in your favour: a large share of the theoretical growth, drawdowns cut to a fraction, and — the real prize — robustness against your own optimism. The tool shows a quarter-Kelly figure beside its picks in that spirit. A discipline, not an instruction.
The bet builder complication
Staking rules assume bets are independent, and a bet builder is the opposite by design. Three legs from the same match rise and fall with the same ninety minutes; sizing each as if it lived alone quietly triples your exposure to one game. Treat the whole builder as a single position, sized once by your combined view of it — and remember from the bet builder guide that the naive combined edge runs optimistic, which argues for sizing below what the headline number suggests, never above.
The rules that outrank every formula
A bankroll is money with no other job. If the honest size of yours is zero, the correct stake is zero and the formulas above are trivia. Keep it in its own account, separate from the money you live on — once the two mix, the limit stops existing. Never stake to recover: the odds don't improve because you're down, and chasing is how small losses become big ones. And if any of this stops being recreation, the responsible gambling page lists free, confidential help. No edge is worth more than that.
Frequently asked
What is the Kelly criterion?
A formula for the stake that maximises long-run bankroll growth if your probability is exactly right: stake the edge divided by the odds minus one. Its power and its danger come from the same place — it takes your edge estimate completely seriously.
Why quarter Kelly instead of full Kelly?
Because your edge is an estimate. Overestimate it and full Kelly doesn't just grow slower, it can turn a winning process into a losing one. A quarter of the stake keeps a worthwhile share of the growth with far shallower drawdowns, and stays sane when the model is a little wrong.
What bankroll do I need to start?
Whatever you could lose entirely without it mattering — money with no other job. If that number is zero, the correct stake is zero. The bankroll's size sets your stakes; it has nothing to do with whether betting is a good idea this month.
Should I increase stakes after a winning run?
Percentage-of-bankroll staking already does this for you, gradually and for the right reason. Doubling up because you feel sharp is how winning runs get given back. The same applies in reverse: never raise stakes to chase a losing run.
Keep going
SharpXI models probabilities; it doesn't promise profit, and neither should anyone else. Betting involves risk — never stake more than you can afford to lose. 18+ — please gamble responsibly.