In one sentence
Fixed staking means putting the same amount on every bet, either a flat £ figure or a fixed percentage of your bank, regardless of odds or recent results.
How it works
Level stakes are the default way to measure whether a strategy works. Because every bet carries the same weight, your profit divided by total staked gives a clean return on investment figure.
There are two common versions. Flat stakes use a fixed £ amount, say £10 a bet, often expressed as one "point". Fixed percentage stakes use, say, 1% of the current bank, so stakes rise after wins and fall after losses, which gives some protection from ruin.
Neither version changes the expected return per £ staked; that comes only from the edge in the prices you take. What staking changes is how much you risk and how bumpy the ride is.
The maths
- n is the number of bets and s the stake per bet in £.
- p is the true win probability, o the decimal odds and c the commission rate.
- σ bet is the standard deviation of the result of a single £1 bet.
In plain English: expected profit grows in line with the number of bets, but the noise grows only with its square root, so the edge shows through slowly.
Worked betting example
You back Over 2.5 goals at 2.10 in 200 matches, £10 flat stakes. Each one truly lands 50% of the time (an illustrative figure), so the fair price is 2.00 and you have an edge.
- Win pays £11, a loss costs £10.
- Expected profit per bet: 0.5 × £11 − 0.5 × £10 = £0.50. Over 200 bets: £100 on £2,000 staked, a 5% ROI.
- Standard deviation of the total: about £148.50. The noise is bigger than the expected profit.
- You lose money if you win 95 bets or fewer. The exact chance of that is 26.2%, so about one in four players with this real edge are behind after 200 bets.
Now add 2% Betfair commission on winnings:
- A win now pays £10.78.
- Expected profit per bet: 0.5 × £10.78 − 0.5 × £10 = £0.39, or £78 over 200 bets (3.9% ROI).
- You now need at least 97 winners to be in profit, and the chance of being behind after 200 bets rises to 31.0%.
A simulation of 200,000 runs of 200 bets gave the same figures to within half a percentage point.
Where it's good
- Evaluating a tipster, model or strategy: ROI at level stakes is the fairest comparison.
- New strategies where you do not yet trust your probability estimates enough to use Kelly.
- Punters who value simplicity and discipline over maximum growth.
- Keeping records that are easy to audit and compare across seasons.
- Fixed percentage staking for a gentle, automatic reduction in risk during losing runs.
Limitations and pitfalls
- Flat stakes ignore the size of your edge, so a 1% edge gets the same money as a 10% edge. That wastes growth when edges vary a lot.
- On mixed odds, level stakes give long shots a much larger influence on the variance of your results than short-priced bets.
- Flat stakes set too large relative to the bank can still lead to ruin; see the risk of ruin page.
- Fixed percentage staking has a drag: a 10% loss followed by a 10% gain leaves you below where you started.
- As the example shows, even real edges need hundreds of bets to show reliably, and commission eats a large share of a small edge.
- Level stakes do not rescue a strategy without an edge. They only lose more predictably.
How to build it
- A spreadsheet is enough: log odds, stake, result and commission for each bet.
- Use scipy.stats.binom to work out the chance of being behind after n bets for your typical odds and edge.
- Many punters set 1 point at 1% to 2% of the starting bank.
- Practical tip: some punters use "level liability" or "stake to win a fixed amount" to even out risk across odds; test which version suits your odds range.
Related methods
- Kelly criterion – sizes stakes by edge instead of fixing them.
- Risk of ruin – how big a flat stake your bank can take.
- Law of large numbers – why the edge only shows over many bets.
- Expected value – the source of any long-run profit.
- Staking progressions – systems that vary stakes after wins and losses, and why they do not help.