Acca Bankroll and Staking Guide: Fixed Stakes and Kelly Caveats

The best staking plan for accumulators is a small, fixed stake that you set before you look at the odds. Accas have long losing runs built into them: a six-leg bet with a 14% chance of winning loses ten times in a row more than one time in five. Staking is about surviving that variance and not turning a bad week into a bad month.

Start with a bankroll, not a stake#

Your bankroll is money set aside for betting that you can afford to lose entirely. It is not your rent, your savings, or money you expect to need next month. Keeping it separate does two jobs: it caps the damage, and it lets you measure what you actually spend.

Then define the stake as a fraction of it.

Fixed stakes#

A fixed stake means the same amount on every acca, or the same small percentage of the bankroll.

Why it works. It removes emotion. You do not chase a loss with a larger stake, and you do not size up because the acca "feels" strong.

The drawdown arithmetic. With a £500 bankroll and a 1% stake that you adjust each time, ten straight losses leave you with £500 × 0.99 to the power of 10 = £452. If you keep the stake fixed at £5, ten losses cost £50 and you have £450. Either way you survive, and that is the point.

How bad can a losing run get?#

Here is the chance of losing a given number of bets in a row at different win rates.

An acca with a 14% chance of winning has a better-than-one-in-twenty chance of going 20 attempts without a win. At an 8% chance (roughly a seven-fold of 70% legs) it is nearly one in five. Plan your stake for runs like these, and keep betting inside the limits you set. If you are tempted to raise stakes to win it back, that is the moment to stop.

The Kelly criterion#

In 1956 John Kelly published a result that connects growth of capital to information in betting (Kelly, 1956, Bell System Technical Journal). The practical form is a staking fraction:

f = (b × p − q) ÷ b

where b is the net odds (decimal odds minus 1), p is your probability of winning, and q is 1 − p.

Example: you estimate 55% on a bet at decimal odds of 2.00, so b = 1. Then f = (1 × 0.55 − 0.45) ÷ 1 = 0.10. Kelly says to stake 10% of the bankroll.

If the edge is zero or negative, f is zero or negative and the formula says do not bet.

Why Kelly is risky with accumulators#

1. Your probability is an estimate. If the true chance is 50%, the same 10% stake has a negative growth rate. The formula is very sensitive to overestimates.
2. Accas multiply the error. If four legs are each really 70% but you rate them at 72%, your combined estimate is 0.72 to the power of 4 = 26.9% against a true 24.0%. A two-point slip per leg becomes a near three-point slip on the bet, and a relative overestimate of about 12%.
3. Edges on accas are usually negative. After compounding margin, most accas have negative expected value. Kelly will tell you to stake zero.
4. Full Kelly is volatile even when you are right. Large swings in the bankroll are part of the strategy.

Here is the first point on a real acca. Take the six-leg bet from our insurance and boost guide: a combined price of 5.36 and an estimated win chance of 13.9% (six legs at 72%).

• Kelly: b = 4.36. f = (4.36 × 0.139 − 0.861) ÷ 4.36 = −5.8%. Do not bet.
• If you had underestimated each leg and the true win chance was 17.4%, Kelly would be −1.5%. Still do not bet.

That is the honest message of Kelly for accas: when margin compounds, you need an exceptionally good edge on every leg before the formula allows a stake at all.

If you use Kelly anyway, bettors commonly stake a fraction, such as half or a quarter, to guard against estimation error. That is a risk-management choice, not a way to create value.

A sensible plan for casual acca bettors#

1. Set a monthly betting budget you can lose entirely.
2. Use fixed stakes of 0.5% to 2% of that budget.
3. Never raise a stake after a loss.
4. Keep a record of every bet, including odds and stake.
5. Take breaks when you feel the need to recover losses.
6. Use the bookmaker's deposit limits and time-outs, and GamStop if you need a longer break.

Does the stake change if the acca is better?#

Only if your estimate of the edge is trustworthy. Starting from a de-vigged market price and a goals model is better than gut feel, as covered in implied probability and removing the margin and how the model works. Even then, error in each leg multiplies through the bet. A tiny edge on four legs can disappear if one leg is wrong.

Do not size up because the price is big. A £2 stake at 40.0 and a £20 stake at 4.0 both aim to win £80, but the second wins roughly ten times as often and loses ten times as much per attempt.

Staking and leg count#

The number of legs sets how long a drought is likely to be. If you want to shorten losing runs, take fewer legs, not bigger stakes. How many legs should an acca have covers the trade-off, and common acca mistakes lists the stake-related traps.

You can build a balanced acca in the ArtificialAcca generator and see how earlier daily accas settled on /results. It is a tool for estimating probabilities, not a way to guarantee profit.

Putting a price on a month of betting#

A fixed stake lets you estimate what a month of betting costs. If a six-leg acca at a 5% margin per leg has an expected return of about 74.6p per £1 (see the hidden cost of accumulators), then a £5 stake costs about £1.27 per bet on average. Twenty such bets in a month cost roughly £25 in expectation, spread across a few wins and many losses.

That is the right way to think about a stake: as the price of an entertainment product with a known running cost. If £25 a month is comfortable and the spending is on the plan you set, the stake is fine. If it is more than you intended, reduce the stake or the number of bets rather than hoping variance will help.

Two practical tools help here:

• A written budget. Decide the monthly amount before the first bet and stop when it is spent.
• Operator limits. Deposit limits and time-outs on your account enforce the plan when you do not feel like it.

Keep a record#

Write down every bet: date, legs, odds, stake and result. After a month, compare your results with what the maths predicts for the odds you took. A six-leg acca at a 14% chance should win about once in seven attempts. If you are well off that either way over a small sample, variance is the most likely explanation, so do not change your stake because of it. A record exists to show you what you spend, not to prove a system.

18+ only. No staking plan makes a negative-expectation bet profitable, and accumulators lose more often than they win. Only gamble with money you can afford to lose. If you are worried about your gambling, contact BeGambleAware.org or the National Gambling Helpline on 0808 8020 133.