Two people can place exactly the same bets over a season and end up with completely different results, because one of them sized the stakes well and the other did not. A staking plan is the rule that decides how much each bet gets. Here are the ones you will meet, and what each one is good and bad at.
Flat staking
The rule: the same amount on every bet — say 1% of your starting bankroll, or a fixed £10.
Flat staking is simple, hard to get wrong, and makes your results easy to judge: if you are up after 300 flat bets, it is because your picks were good, not because a few big stakes landed. Its weakness is that it treats every bet the same, whether the edge is large or small, and it does not compound. It is the right place for most people to start.
Percentage staking
The rule: a fixed percentage of your current bankroll — 2%, for example.
Stakes grow as you win and shrink as you lose, so a losing run cannot take you to zero. Your bankroll compounds. It is still simple, and it is what many professionals use day to day. The drawback is the same as flat staking: it ignores how good each bet is.
Kelly and fractional Kelly
The rule: stake in proportion to your edge — (odds × probability − 1) ÷ (odds − 1) — then take a fraction of that, usually a half or a quarter.
Kelly is the only method on this list that uses your estimate of how good the bet is. Bigger edge, bigger stake; no edge, no bet. Used at full strength it is the fastest way to grow a bankroll and also the most stressful, with deep drawdowns. At half or quarter strength it keeps most of the growth and loses most of the stress. The price of admission is that you must be able to estimate probabilities honestly. We explain the maths in What is the Kelly criterion?
Level profit (“stake to win”)
The rule: stake whatever is needed to win a fixed amount — £100 profit on every bet, whatever the odds.
This means large stakes on short prices and small stakes on long ones, so your results are not dominated by a few big-priced winners. Useful for tidy accounting; it still ignores your edge.
Martingale — do not
The rule: double your stake after every loss, so the next win recovers everything.
Martingale feels safe because it wins small amounts often. It fails because losing runs are normal and stakes grow very fast. Ten losses in a row at even money — which happens about once in every thousand runs — needs a stake 1,024 times your first one, and a bankroll of over 2,000 units to get there. Every Martingale bettor eventually meets a run they cannot fund, and that one run costs more than all the small wins put together. No staking plan can turn a bet without an edge into a winner; Martingale just hides the loss until it arrives all at once.
Fibonacci, D’Alembert and other “progressions”
These are gentler versions of the same idea: raise stakes after losses, lower them after wins. They go broke more slowly than Martingale, but for the same reason. If you find yourself raising a stake because you lost the last one, stop.
Side by side
| Method | Uses your edge? | Compounds? | Risk of ruin | Best for |
|---|---|---|---|---|
| Flat | No | No | Low | Beginners, testing a strategy |
| Percentage | No | Yes | Very low | Steady long-term betting |
| Fractional Kelly | Yes | Yes | Very low | Model-driven betting with a trusted edge |
| Full Kelly | Yes | Yes | Low, but huge swings | Almost nobody |
| Martingale & progressions | No | No | Certain, eventually | Nobody |
Which one should you use?
Start flat, at 1–2% of your bankroll, until you have a few hundred bets of evidence that your selections beat the market. Then move to percentage staking so your bankroll compounds. Move to fractional Kelly only when you have a model that produces probabilities you have tested and trust. And never, at any stage, raise a stake because the last one lost.