A staking plan fixes how the size of a stake is determined at each decision — never whether to bet. Flat stakes keep a constant value; percentage of bankroll and fractional Kelly (see Kelly criterion) adjust it to the current capital; negative progressions (martingale, d'Alembert) increase exposure after a loss without changing the expected value of the next bet. Example: on a €1,000 bankroll, a 1% flat stake means €10 on every bet, whether it follows a winning or losing run — which is exactly what makes it predictable.
What a staking plan fixes, and what it doesn't
A staking plan sets the size of a stake at each decision — never whether to bet, or on which outcome. It creates no positive expected value and corrects no wrong estimate: it only changes how a capital absorbs a run of results, favourable or not.
That distinction frames the whole page: every method below is judged on its effect on risk of ruin and drawdown, never on a return none of them can guarantee.
A staking plan doesn't make profitable any method that wasn't already. It only redistributes exposure over time; it never touches the expected value of the next bet.
Flat stakes: the reasonable default
Flat staking keeps the same absolute value on every bet, regardless of the bankroll's balance. On a €1,000 bankroll, a 1% flat stake means €10 on every decision — ahead of a winning run just as much as a losing one.
Its strength is simplicity: exposure per bet stays predictable and easy to track over time (see tracking your bets). Its limit is the mirror image: it doesn't automatically reduce exposure after a losing run, which requires manual discipline to adjust the stake if the bankroll drops for a sustained period.
Fixed percentage of bankroll
A fixed percentage recalculates the stake at each decision on the current balance of your bankroll, not on its starting amount. A 1% stake is worth €10 on €1,000, then €9.50 if the bankroll drops to €950 — it shrinks mechanically after a loss, and grows mechanically after a win.
This mechanic slows how fast an unfavourable run can exhaust a capital, compared with a flat stake that stays identical regardless of the balance. In exchange, it produces a stake that varies from bet to bet, less simple to track than a fixed amount.
Kelly and fractional Kelly
The Kelly criterion pushes the percentage-of-bankroll principle further: the fraction is no longer arbitrary, it follows from odds and an estimated theoretical edge, not a percentage picked at random.
The formula, its sensitivity to estimation error, and the f* ≤ 0 case →
Its trade-off is a high sensitivity to the quality of that probability — the one variable the formula can't measure on its own. That's why only its fractional version (half or a quarter of the calculated fraction) is actually used in practice: it dampens the effect of an overly optimistic estimate without cancelling out the point of the formula.
Negative progressions: martingale and d'Alembert
A negative progression increases the stake after a loss, with the idea of recovering it on the next win. The martingale doubles the stake after every loss; d'Alembert raises it by a fixed step. Both change only exposure — never the expected value of the next bet, which stays exactly what it would have been without the progression.
"The martingale always wins eventually."
A martingale does recover a loss on the next win — as long as the bankroll and the operator's stake limits allow it. The problem isn't the recovery itself: it's that exposure grows exponentially over a losing run, until it exceeds the available bankroll or the stake limit well before a win comes along. See risk of ruin for what an exposure that grows this way produces.
This mechanic is consistent with what the tipsters page already shows: a martingale marketed as risk-free doubles exposure without ever changing a bet's real edge.
Positive progressions: the paroli
The paroli does the opposite of a martingale: the stake grows after a win, never after a loss. The extra exposure then comes from profit already banked, not from additional capital committed — the maximum loss on a losing run stays capped at the starting stake.
The paroli doesn't escape the rule shared by every staking plan, though: it creates no positive expected value. It only changes the speed at which a profit builds up, or disappears, over a run of consecutive wins.
One capital, five ways to expose it
On a €1,000 bankroll and a run of three consecutive losses, the five plans don't reduce the capital the same way.
- Flat stake at 1% (€10): three losses take out €30, the bankroll drops to €970.
- Percentage at 1%: the third stake, recalculated on €980, is worth €9.80 — the gap with flat staking stays small over a short run, but builds up over a longer one.
- Fractional Kelly: the stake depends on the estimated edge at each bet, so on context — it can rise or fall independently of the balance alone.
- Martingale starting at €10: successive stakes run €10, €20, then €40 — €70 committed on the third loss alone, seven times the starting stake.
- d'Alembert starting at €10: successive stakes run €10, €20, €30 — linear growth, not exponential, but still unrelated to the expected value of the next bet.
- Paroli starting at €10: the stake stays at €10 as long as no win is recorded — exposure on a losing run never exceeds the starting stake.
The difference doesn't show on a single bet. It shows in how fast each method exposes an identical capital to an identical unfavourable run.
Comparing the plans
| Staking plan | Effect on exposure | Effect on risk of ruin |
|---|---|---|
| Flat stake | Constant, independent of the balance | Predictable; doesn't adjust itself after a loss |
| Percentage of bankroll | Shrinks after a loss, grows after a win | Mechanically slows how fast capital is exhausted |
| Fractional Kelly | Derived from an estimated edge, deliberately reduced | Depends entirely on the quality of the estimate |
| Martingale / d'Alembert | Grows after a loss, sometimes exponentially | High: exposure can exceed the bankroll or stake limit |
| Paroli | Grows after a win, from profit already banked | Capped at the starting stake on a losing run |
What no staking plan does
None of the plans above corrects a wrong estimate, or turns a negative expected value into a positive one. They only organise exposure over time — which the page on risk management confirms, drawing the exact line between what's controllable and what isn't.
Responsible gambling. No staking plan guarantees a positive performance. A stake can be lost in full, whatever sizing method is used. Never commit money to betting that you need for daily life, housing, bills or your emergency savings, and use the deposit limits or self-exclusion tools offered by licensed operators. Learn more about responsible gambling.
Dig into the market
Odds movements are only part of the story. Here are the next topics to read.
Profitability and risk: the complete guide
The full picture: profit, ROI, expected value, variance, sample size, drawdown, risk.
Read the guideKelly criterion: what the formula sizes
The formula, its sensitivity to estimation error, and the f* ≤ 0 case.
Understand the Kelly criterionBankroll in sports betting
The capital every staking plan applies to.
Understand bankrollRisk of ruin: exhausting a bankroll
What a staking plan actually changes, seen in reverse.
Understand risk of ruinDrawdown: measuring the dips in a bankroll
The other measure a staking plan should be judged against.
Understand drawdownTracking your bets
The data needed to check what a staking plan actually produced.
See what to recordFrequently asked questions
What is a staking plan?
It's the rule that sets the size of a stake at each decision — a fixed amount, a share of the bankroll, or a progression tied to previous results. It never says whether to bet, or on what outcome.
What's the most reasonable default staking plan?
Flat stakes: a constant amount, independent of the bankroll's balance. It stays predictable and easy to track over time, at the cost of not adjusting itself after a losing run.
Is percentage of bankroll safer than flat stakes?
It mechanically slows how fast a capital can be exhausted, because the stake shrinks with the balance after a loss. In exchange, it produces an amount that varies from bet to bet, less simple to track than a flat stake.
Why is the martingale discouraged?
Because it grows exposure exponentially after every loss, without ever changing the expected value of the next bet. Exposure ends up exceeding the available bankroll or the operator's stake limit long before a win comes along.
Is the Kelly criterion a staking plan like any other?
It's a percentage of bankroll derived from odds and an estimated probability, rather than chosen arbitrarily. Its trade-off is a high sensitivity to the quality of that probability, covered in detail on its dedicated page.
What is a positive progression like the paroli?
The stake grows after a win, never after a loss: the extra exposure comes from profit already banked, not from additional capital. The maximum loss on a losing run stays capped at the starting stake.
Can a staking plan make a strategy profitable?
No. No staking plan creates a positive expected value or corrects a wrong estimate. Each one only organises exposure over time, which the page on risk management confirms in detail.
Sources & methodology
This page compares common staking plans using Kelly's founding paper on optimal sizing of a repeated stake and Thorp's work on favourable gambling systems, together with the odds-analysis methodology developed by OddScore.
- Judge each plan on its effect on risk of ruin and drawdown, never on a return none of them can promise.
- Treat negative progressions (martingale, d'Alembert) as an explicit debunk: they change exposure, never the expected value of the next bet.
- Route the detail of the Kelly criterion to its dedicated page rather than duplicating the formula here.
- Illustrate every plan on the same bankroll and stake, to keep the comparison readable.