Bankroll Management: How to Protect Your Betting Money
Flat staking, percentage of bank and the Kelly criterion run over the same ten bets, with every figure in naira — plus drawdown maths and the honest limits of any staking plan.
Bankroll management is the set of rules that decides how much of your money goes on any single bet, and it is the one part of betting you control completely. Get it wrong and an ordinary run of ten losses ends your season; get it right and the same run costs you a fifth of your money and nothing more. Below, the three staking plans punters actually use — flat stakes, a percentage of the bank and the Kelly criterion — are run over one identical set of ten bets in naira, followed by an honest look at variance and at the part no staking plan can fix.
In short
- A bankroll is money set aside for betting only. If losing all of it changes your month, it is too big.
- The 1–3% rule is arithmetic, not folklore: ten straight losses cost about a fifth of the bank at a 2% stake and two thirds of it at a 10% stake.
- On the same ten bets, flat staking finished at ₦100,800, a 2% plan at ₦100,577 and full Kelly at ₦105,761 — but Kelly's worst dip was 13.8% against 4.5% for flat.
- Kelly is only as good as your own probability estimate. Overstating a chance by five percentage points turns a growing bank into a shrinking one.
- No staking plan turns a negative edge positive. It changes the speed and the shape of the loss, never the sign.
What a bankroll actually is
A bankroll is a fixed sum, kept separate from the money you live on, that exists to absorb losing runs. It is not your salary, not next month's rent and not borrowed money. The practical test is simple: if losing the whole amount over the next three months would force you to change a single real plan, the amount is too large.
Two numbers define it. The bank is the total. The unit is one standard stake, written as a percentage of the bank. Everything else in this guide is a rule for turning the bank into the unit. If you are still working out the basics of a bet slip, start with our guide to sports betting for beginners and come back to this page afterwards.
Top the bank up on a schedule you decide in advance — say, once a month — never in the middle of a bad evening. A top-up made after a loss is not bankroll management; it is the loss deciding your budget for you.
Where the 1–3% rule comes from
The rule is not a superstition, it is what survives a losing run. Assume ten losses in a row from a ₦100,000 bank and compare stake sizes. Flat staking loses stake × 10; a percentage plan compounds downwards, so the damage is slightly smaller in naira but far larger as the percentage grows.
| Stake per bet | Bank after 10 straight losses | Share of the bank gone | Gain needed to get back |
|---|---|---|---|
| 2% of the bank | ₦81,707 | 18.3% | +22.4% |
| 5% of the bank | ₦59,874 | 40.1% | +67.0% |
| 10% of the bank | ₦34,868 | 65.1% | +186.8% |
| 20% of the bank | ₦10,737 | 89.3% | +831.4% |
Ten losses in a row is not an exotic event. A punter who wins 45% of his bets loses 55% of them, so any given six-bet stretch is a clean sweep of losses with probability 0.55⁶ = 2.8%. Across 100 bets there are 95 possible starting points for such a stretch, which works out at roughly two or three of them per hundred bets. The plan has to be built for that, not for the good weeks.
The three plans, defined
- Flat staking. The same naira amount on every bet, recalculated only when you formally resize the bank. A 2% unit on ₦100,000 is ₦2,000, and it stays ₦2,000 whatever happens next.
- Percentage of bank. A fixed share of the current bank, recalculated before every bet. Stakes shrink automatically in a drawdown and grow as the bank grows.
- Kelly criterion. The stake is proportional to your edge. The fraction of the bank to stake is f = (p × odds − 1) ÷ (odds − 1), where p is your own estimate of the chance. No estimate, no Kelly. Many people stake half the number the formula gives; that variant is called half Kelly.
Kelly needs a probability, so it needs you to convert prices into percentages first — the mechanics are in our guide on how to read betting odds.
The same ten bets under all three plans
Here is the sample. Every bet has an estimated chance above the one implied by the price, because a staking plan has nothing to work with otherwise. Treat the edges below as a demonstration, not as a claim: real edges are much smaller, much rarer and never sit on every bet in a row.
| # | Odds | Implied by the price | Your estimate | Edge (p × odds − 1) | Result |
|---|---|---|---|---|---|
| 1 | 2.10 | 47.6% | 52% | +9.2% | Won |
| 2 | 1.85 | 54.1% | 58% | +7.3% | Lost |
| 3 | 2.40 | 41.7% | 45% | +8.0% | Lost |
| 4 | 1.95 | 51.3% | 55% | +7.3% | Won |
| 5 | 3.10 | 32.3% | 35% | +8.5% | Lost |
| 6 | 1.75 | 57.1% | 62% | +8.5% | Won |
| 7 | 2.25 | 44.4% | 48% | +8.0% | Lost |
| 8 | 2.00 | 50.0% | 53% | +6.0% | Won |
| 9 | 1.90 | 52.6% | 56% | +6.4% | Lost |
| 10 | 2.60 | 38.5% | 42% | +9.2% | Won |
Five winners, five losers, a starting bank of ₦100,000 and a 2% unit for the first two plans. The stakes each plan produces look like this (rounded to the naira).
| # | Flat, ₦2,000 | 2% of current bank | Full Kelly | Half Kelly |
|---|---|---|---|---|
| 1 | 2,000 | 2,000 | 8,364 (8.4%) | 4,182 |
| 2 | 2,000 | 2,044 | 9,378 (8.6%) | 4,492 |
| 3 | 2,000 | 2,003 | 5,704 (5.7%) | 2,860 |
| 4 | 2,000 | 1,963 | 7,183 (7.6%) | 3,711 |
| 5 | 2,000 | 2,000 | 4,086 (4.1%) | 2,039 |
| 6 | 2,000 | 1,960 | 10,977 (11.3%) | 5,595 |
| 7 | 2,000 | 1,990 | 6,726 (6.4%) | 3,294 |
| 8 | 2,000 | 1,950 | 5,902 (6.0%) | 2,989 |
| 9 | 2,000 | 1,989 | 7,414 (7.1%) | 3,649 |
| 10 | 2,000 | 1,949 | 5,569 (5.8%) | 2,846 |
Example Bet 6, odds 1.75, your estimate 62%. Kelly fraction = (0.62 × 1.75 − 1) ÷ 0.75 = 0.085 ÷ 0.75 = 11.3% of the bank. At that point the Kelly bank stood at ₦96,856, so the stake was ₦10,977 and the winning return ₦19,210 (profit ₦8,233). The flat plan staked ₦2,000 on the identical bet and made ₦1,500. Same opinion, same price, five and a half times the money at risk.
After all ten bets:
| Plan | Final bank | Profit | Return on the bank | Deepest dip |
|---|---|---|---|---|
| Flat ₦2,000 | ₦100,800 | +₦800 | +0.8% | −4.5% |
| 2% of bank | ₦100,577 | +₦577 | +0.6% | −4.6% |
| Full Kelly | ₦105,761 | +₦5,761 | +5.8% | −13.8% |
| Half Kelly | ₦103,530 | +₦3,530 | +3.5% | −7.0% |
Three things are worth reading off that table. First, Kelly earned the most because it staked the most — and its worst moment was three times deeper than flat's. Second, the percentage plan finished behind flat staking despite an identical unit, because the bank was below its starting point for most of the sequence, so the winners were backed with smaller stakes than the losers. That is the built-in cost of a percentage plan, and the price of its main benefit: it cannot run the bank to zero. Third, all of it rests on the estimates in column four being honest.
Same record, different money
Now keep the plans, the prices and the estimates exactly as they are, and simply move the luck: let the five bets that lost be the five that win instead. The record is still five and five.
| Plan | Original sequence | Luck reversed | Swing |
|---|---|---|---|
| Flat ₦2,000 | +₦800 | +₦3,000 | ₦2,200 |
| 2% of bank | +₦577 | +₦2,751 | ₦2,174 |
| Full Kelly | +₦5,761 | −₦3,933 | ₦9,694 |
| Half Kelly | +₦3,530 | −₦1,209 | ₦4,739 |
The same strike rate, the same prices and the same rules produce a ₦9,694 swing under Kelly and a ₦2,200 swing under flat stakes. That gap is variance, and it is the reason ten bets — or fifty, or a hundred — tell you almost nothing about whether a method works. Big stakes on the wrong five games look identical to bad judgement until the sample gets long enough to separate them. Long accumulators magnify the same effect, which is covered in the guide to singles, accumulators and system bets.
Kelly's weak point is your own number
Kelly assumes p is correct. It is unusually harsh when it is not, because the stake grows with the claimed edge — and an inflated edge is exactly what an over-confident estimate produces.
Important A bet is priced at 2.10 and its true chance is 50%. You estimate 55%. Kelly on your estimate says stake (0.55 × 2.10 − 1) ÷ 1.10 = 14.1% of the bank; the correct figure is (0.50 × 2.10 − 1) ÷ 1.10 = 4.5%. Repeat that bet 100 times at 14.1% and the bank's expected growth factor is 0.68 — a third of it is gone — while staking the correct 4.5% grows it by a factor of 1.12. A five-point error in your head, and the "optimal" plan is the one destroying the account.
This is why fractional Kelly exists. Staking half or a quarter of the formula's output keeps most of the growth and cuts the damage from bad estimates, which is a reasonable trade when the estimates come from a human rather than a tested model.
What a staking plan cannot do
A staking plan distributes risk. It does not create an edge, and this is the single most important sentence on the page.
The price already contains the bookmaker's margin. Take a two-way market at 1.90 on each side. The implied chances are 1 ÷ 1.90 = 52.6% each, and they add up to 105.3% rather than 100%. That 5.3 points of overround is a margin of 1 − (1 ÷ 1.053) = 5.0% on turnover, and it is charged whichever side you back.
Negative expectation stays negative. Stake ₦1,000 at odds of 2.00 on something whose real chance is 47.5%. The expected value per bet is (0.475 × ₦1,000) − (0.525 × ₦1,000) = −₦50. Over 200 bets that is −₦10,000, and no staking rule changes it: flat staking loses it slowly and evenly, a percentage plan loses it a little more slowly as the bank shrinks, and Kelly refuses to stake at all because the formula returns a negative fraction. What a plan controls is how long you last and how violently the balance moves — not the sign of the result.
Progressive recovery systems make it worse. Doubling after each loss from ₦1,000 at odds of 2.00 puts ₦63,000 through the account by the sixth step to win the original ₦1,000, and six losses in a row at even money happen 1.6% of the time — often enough that anyone betting regularly will meet it. Maximum-stake limits then stop the sequence before it can "correct". Chasing sits at the top of our list of beginner betting mistakes for that reason.
Drawdowns: the arithmetic of getting back
Losses and recoveries are not symmetrical, because the recovery is calculated on a smaller bank.
| Bank is down | ₦100,000 becomes | Gain needed to be level |
|---|---|---|
| 10% | ₦90,000 | +11.1% |
| 20% | ₦80,000 | +25.0% |
| 30% | ₦70,000 | +42.9% |
| 50% | ₦50,000 | +100.0% |
| 70% | ₦30,000 | +233.3% |
Everything to the left of 30% is recoverable with ordinary betting. Everything to the right needs stakes you should not be taking, which is how a drawdown becomes a wipe-out. Set a stop rule before you need one: a fixed number of units per day, and a monthly point at which you stop and review rather than reload.
Setting your own plan up in six steps
- Fix the bank. Money you can lose without changing any plan. A ₦20,000 bank is a perfectly serious starting point.
- Fix the unit. 1–3% of the bank. On ₦20,000 at 2% that is ₦400 per bet — and yes, that is meant to feel small.
- Fix the daily and weekly caps. For example, no more than three units in a day and no top-ups within the month. The exact numbers are a convention; having them in writing before kick-off is not.
- Pick one plan and leave it alone. Flat staking is the default and the benchmark. Move to a percentage plan if you want automatic shrinkage in a drawdown, and to half Kelly only if you genuinely produce your own probabilities.
- Log every bet. Date, match, market, odds, stake, result, profit. Without a log you are remembering, and memory edits losses out.
- Review monthly. Return on turnover = profit ÷ total staked. The flat sequence above made ₦800 from ₦20,000 staked, which is 4.0% — a good month, and far too short a sample to prove anything.
Resize the bank only at review time, up or down, and never mid-session. Our predictions for today and the match centre are there to help you choose selections; the sizing of them stays on this page.
Mistakes that undo a good plan
- Raising the stake because a bet "feels certain" — confidence is not an input to any of the three formulas.
- Rounding the unit up after a win and down after a loss, which is a percentage plan running backwards.
- Counting an accumulator's ₦2,000 stake as one unit while it carries the variance of five bets.
- Reloading the account mid-evening to "finish level".
- Keeping betting money in the same wallet as everyday money, so the bank has no edges and no size.
FAQ
How much should I stake per game in naira?
Between 1% and 3% of your betting bank, decided before you look at any fixture. On a ₦20,000 bank that is ₦200–₦600 per bet; on ₦100,000 it is ₦1,000–₦3,000. The naira figure matters less than the fact that it is a fixed share of a bank you have defined, and that ten consecutive losses would still leave you with most of it.
Is the Kelly criterion better than flat staking?
Only if your probability estimates are genuinely accurate. Kelly maximises long-run growth when p is right, and it overstakes badly when p is inflated — a five-point error can turn a growing bank into one that shrinks by a third over 100 bets. Flat staking needs no estimate at all, which is why it remains the default and the benchmark other plans are measured against.
Should I raise my stakes after a winning run?
Not because of the run itself. Past results do not change the price of the next match, so a streak is not information about it. Stakes rise when the bank rises under a percentage or Kelly plan, or when you formally resize the unit at a scheduled review — never in the middle of a session because things are going well.
How large should a bankroll be before I start?
Large enough that one unit is a sum you can bet without thinking about it, and small enough that losing the lot changes nothing in your budget. Most people work backwards: take the monthly amount you can afford to lose, treat it as the bank, then divide by 50 to get a unit.
Read next
- Sports Betting for Beginners — how a bet, a slip and a settlement actually work.
- 15 mistakes beginners make — the habits that break a staking plan.
- Responsible gambling — limits, time-outs and where to get help.
Disclaimer This article is for information only and is not an inducement to gamble. Betting involves the risk of losing money — never stake more than you can afford to lose. 18+. If gambling stops being entertainment, read our responsible gambling guide and seek help.
The editorial prediction is provided for informational purposes only and should not be treated as a direct call to action. Every reader should do their own analysis before deciding on a bet.