You can be brilliant at finding value bets and still lose money. This happens all the time. The punter who has a genuine edge but stakes wildly — £200 on a whim one day, £20 the next, £500 when they’re “feeling confident” — will often end up worse off than a mediocre punter who manages their money properly.
Staking and bank management are boring. Nobody got into racing because they were excited about proportional staking. But they’re the difference between a hobby that occasionally pays its way and one that bleeds money.
What a betting bank is
A betting bank is a set amount of money dedicated to betting. It’s separate from your living expenses, your savings, your bill money. It’s the amount you can afford to lose entirely without it affecting your life.
That last part is important. If losing your entire betting bank would cause genuine financial stress, the bank is too big. Set it at a level where, if the worst happened and you lost the lot, you’d be annoyed but not in trouble.
A common starting point is 50-100 “points.” Each point is one unit of your standard bet. If your bank is £500 and you set it at 100 points, each point is £5. Your standard bet is £5, or one point. Some bets might be 2 points (£10) if you’re particularly confident. You rarely go above 3 points on any single bet.
The point system does two things. It keeps you disciplined because you’re thinking in units rather than pounds, which helps reduce emotional betting. And it gives you a framework that adjusts if your bank grows or shrinks.
Level stakes
The simplest staking plan. Every bet is the same size: one point. You back a 2/1 shot? One point. You back a 20/1 outsider? One point.
The advantage is simplicity. You don’t need to calculate anything. You don’t need to decide how confident you are. Every bet is treated equally.
The disadvantage is that it ignores your confidence levels. If you genuinely think a 5/1 shot is the bet of the week and a 12/1 shot is a speculative flyer, level stakes treats them the same. You’re putting the same money on a strong opinion as a weak one.
Despite that, level stakes is a perfectly good system. It removes the temptation to overbet, which is most punters’ biggest leak. Professional tipsters almost always quote their results at level stakes because it’s the only fair baseline — if a tipster can’t show a profit at level stakes, their staking is just adding risk, not skill.
If you’re unsure which staking plan to use, start with level stakes. You can always move to something more sophisticated later.
Proportional staking
With proportional staking, your bet size adjusts as your bank changes. If your bank is £500 and you stake 2% per bet, your first bet is £10. If you win and your bank grows to £550, your next bet is £11. If you lose and your bank drops to £450, your next bet is £9.
The idea is that you increase stakes when you’re winning (to maximise growth) and decrease when you’re losing (to protect the bank). It’s mathematically sound and it’s very hard to go completely broke because your stakes keep shrinking as the bank declines. You’d need an absurdly long losing streak to get from a meaningful bank to zero.
The standard range for proportional staking is 1-3% of your bank per bet. Conservative punters lean towards 1%. More aggressive ones might go to 3% or occasionally 5% for strong bets.
The practical downside is that you need to recalculate your stake before every bet, which is mildly annoying. And your stakes are always going down after losses, which means you need to win at higher odds just to get back to where you started. Psychologically, it can be frustrating to see your bet size shrink during a losing run, even though that’s exactly what the system is designed to do.
The Kelly Criterion
The Kelly Criterion is the mathematically optimal staking plan for maximising long-term growth. It tells you exactly what fraction of your bank to bet based on your edge and the odds.
The formula is:
Stake = (bp - q) / b
Where:
- b = the decimal odds minus 1 (so for 4/1, b = 4)
- p = your estimated probability of winning
- q = the probability of losing (1 - p)
An example. You think a horse has a 25% chance of winning (p = 0.25, q = 0.75). The odds are 5/1 (b = 5).
Stake = (5 × 0.25 - 0.75) / 5 = (1.25 - 0.75) / 5 = 0.50 / 5 = 0.10
Kelly says to stake 10% of your bank.
If you think the same horse has a 20% chance at 5/1:
Stake = (5 × 0.20 - 0.80) / 5 = (1.0 - 0.80) / 5 = 0.20 / 5 = 0.04
Kelly says 4% of your bank. The edge is smaller, so the stake is smaller. That makes intuitive sense.
Kelly has two big problems in practice.
You need accurate probability estimates. The formula is only as good as your estimate of p. If you think a horse has a 25% chance but it really has a 15% chance, Kelly will tell you to overbet massively. Since nobody’s probability estimates are perfect, full Kelly staking is extremely volatile. Professional gamblers who use Kelly typically stake at half-Kelly or quarter-Kelly to reduce the variance.
It produces large stakes on high-edge bets. If you find a horse you think has a 40% chance at 5/1, Kelly suggests staking 28% of your bank on a single bet. That’s terrifying, and one losing bet at that level would devastate your bank. Again, fractional Kelly (half or quarter of the recommended stake) brings this down to something manageable.
The practical takeaway: Kelly is a good conceptual framework. It tells you to bet more when your edge is bigger and less when it’s smaller, which is obviously correct. But using it at full strength requires probability estimates you almost certainly don’t have. Half-Kelly or quarter-Kelly is a reasonable compromise.
What about rating-based staking?
Some punters stake according to their confidence: 1 point for a standard bet, 2 points for a strong bet, 3 points for the bet of the day. This is basically level stakes with an adjustment for confidence.
It works if your confidence levels are calibrated. If your 3-point bets genuinely win more often (or at higher odds) than your 1-point bets, the weighting adds value. But many punters find that their “strong” bets don’t actually perform better — the confidence is emotional rather than analytical. If you’re going to use a rating system, track the results of each confidence level separately. If your 3-point bets show a worse ROI than your 1-point bets, your confidence rating is doing more harm than good and you should revert to level stakes.
Losing runs and variance
No matter which staking plan you use, you will have losing runs. Not a possibility — a certainty. Even with a genuine 10% edge (which is extremely good), a losing run of 15-20 bets is statistically likely within any 500-bet sample.
With level stakes at 1 point per bet, a 20-bet losing run costs you 20 points. If your bank is 100 points, that’s a 20% drawdown. Unpleasant but survivable.
With proportional staking at 3% per bet, a 20-bet losing run costs less in absolute terms because the stakes decrease as the bank shrinks, but the percentage drawdown is still significant.
The point is that your bank needs to be large enough to absorb losing runs. If you’re staking 5% of your bank per bet, a bad week can wipe out 30% or more. If you’re staking 1%, the same bad week costs you 7-8%. The smaller the stakes relative to the bank, the smoother the ride.
A 100-point bank with level stakes at 1 point per bet is the standard recommendation for a reason. It gives you enough runway to survive the inevitable bad stretches without needing to top up the bank.
Rules for bank management
Never chase losses. After a losing day, the temptation is to put a bigger bet on the next race to “get it back.” This is the single most destructive behaviour in betting. If your staking plan says one point, it’s one point, whether you’re 30 points up or 30 points down. No exceptions.
Review your bank monthly. If it’s grown, you can increase your point size (though you don’t have to — banking some profit is sensible). If it’s shrunk, reduce your point size. If it’s down more than 30%, take a break and review your selections before continuing.
Don’t mix bank money with general funds. Keep your betting bank in a separate account or at least track it separately. The moment you start “borrowing” from the bank or topping it up from your current account, you’ve lost control of the process.
Set a stop-loss. Decide in advance how much you’re willing to lose before you stop and reassess. A reasonable stop-loss is 50% of your starting bank. If your 100-point bank drops to 50 points, stop betting for a week. Review your recent bets. Are you making bad selections, or is it just variance? If it’s variance, you can resume. If your approach is flawed, fix it before risking more money.
Withdraw profits periodically. If your bank grows from 100 to 150 points, take 25 points out. Put it in your savings. Enjoy it. Reinvesting every penny back into the bank feels disciplined but it also means you never actually benefit from your betting. The purpose of the bank is to make money, not to accumulate an ever-larger number on a bookmaker’s website.
Which plan should you use?
If you’re starting out or you want simplicity: level stakes. One point per bet, 100-point bank. It works, it’s easy, and it removes the biggest source of error (emotional staking).
If you want to optimise: proportional staking at 1-2% per bet. This gives you the benefits of automatic adjustment without the volatility of Kelly.
If you’re experienced and confident in your probability estimates: half-Kelly or quarter-Kelly. This maximises long-term growth while keeping the variance manageable.
And if someone tries to sell you a staking plan that “guarantees” profits — a system where the staking itself generates returns regardless of selection quality — walk away. No staking plan can turn losing selections into profit. Good staking can maximise the returns from good selections and minimise the damage from bad ones. That’s all it can do, and that’s enough.