Staking without going broke
A method with a real edge and stakes that are too large still ends at zero. Staking decides whether the edge ever gets a chance to arrive.
On this page
Two people use the same method, find the same bets at the same prices, and one of them ends the year up and the other ends it at zero. The difference is not selection. It is how much they put on.
Why the arithmetic is unforgiving
A bankroll that loses 50% needs to gain 100% to recover. That recovery asymmetry is the whole problem: a deeper drawdown requires a progressively larger percentage gain to get back to the starting point.
Now recall what variance does to a good bet: under independent outcomes, a method winning 53% at even money has about a 47% chance of at least one run of eight losses in 500 bets, and about a 25% chance of a run of nine. Stake 10% of the current bankroll per bet and nine losses leave about 39% of the starting bankroll. That run alone would not show that the method was wrong, but it would show how exposed the staking rule was.
The floor nobody can lower
With exact proportional staking and no minimum bet, a bankroll can approach zero without mathematically reaching it. Real accounts have minimum stakes and people have practical stopping thresholds, so a severe drawdown can still make the plan unusable. Stake size is a major control, but the edge, odds, dependence between bets and stopping rule matter too.
This is why a stated edge should make you more careful rather than less. An estimated 6% edge that is really 1% still makes money slowly; the same estimate staked as though it were certain does not.
Where uncertainty belongs
Every edge published here is an estimate with error around it. Research on Kelly betting under parameter uncertainty finds that shrinking the stake can improve out-of-sample performance when the win probability is estimated rather than known. The more uncertain the estimate, the less confidence there is in a large stake.
What this does not tell you
No staking scheme creates an edge. Applied to bets with negative expectation, careful staking makes you lose more slowly and does not make you win. Staking is a way of surviving long enough for an edge to matter, and if there is no edge it is a way of surviving to lose.
Common questions
What is a sensible stake size?
There is no universal percentage. It must fit a loss limit set from money you can afford to lose, the prices you bet, the uncertainty in your edge and the drawdown you can tolerate.
Does a bigger edge justify a bigger stake?
Yes, in proportion, but only if the edge is real. Staking up on an estimated edge that turns out to be zero loses money faster than staking flat would.
What is risk of ruin?
The probability that losses take a bankroll to zero or below a usable threshold. It depends on the edge, odds, staking rule and stopping threshold, and generally rises as stakes get larger.