Betting strategies8 min read
Betting Bankroll Management: Units, Staking Plans and Surviving Bad Runs
Betting bankroll management means setting aside a fixed sum you can afford to lose, splitting it into small units of about 1–2% each, and staking a consistent amount per bet instead of whatever feels right. It can't turn losing bets into winning ones, but it stops a normal bad run from wiping you out.
Key takeaways
- Your bankroll is money you can lose without touching rent, bills or savings. If that amount is zero, the right stake is zero.
- One unit of 1–2% of your bankroll per bet is the sensible default. At 5% or more, an ordinary losing streak does serious damage.
- Flat staking is the simplest and hardest to mess up. Percentage staking adjusts itself as your bankroll moves.
- The Kelly criterion sizes stakes by your edge, but it's only as good as your probability estimate, so most people who use it bet a half or a quarter of what it says.
- No staking plan creates an edge. Staking decides how long you last, not whether your bets are good.
Nobody opens a betting account because they’re excited about stake sizing. But betting bankroll management is the one skill that helps every bettor, from the person having a weekend flutter to the spreadsheet obsessive, because it’s the only part of betting you fully control. You can’t control whether a striker misses from two metres. You can control how much that miss costs you.
I’ll keep this practical: what a bankroll is, how big a unit should be, the main staking plans with numbers, a short look at Kelly, and what a normal losing run looks like so it doesn’t scare you into doing something daft.
What a bankroll actually is
A bankroll is a fixed pot of money set aside only for betting. Not your current account. Not “whatever’s left at the end of the month”. A number you decide in advance, that you could lose entirely without consequences beyond being annoyed.
Two rules make the rest of this guide work:
- It’s separate. Ideally it’s physically separate: its own e-wallet or a bookmaker balance you don’t top up on impulse.
- It’s finite. If it goes, you stop. You don’t reload the same evening. Set a deposit limit with your bookmaker (or sportsbook, in US terms) to enforce that. Licensed operators in most regulated markets offer them.
If you can’t name an amount you’re comfortable losing, the correct bankroll is zero and the correct hobby is watching the match.
Units: the simplest betting bankroll management tool
A unit is your standard stake, defined as a percentage of your bankroll. It turns “how much should I put on this?” from an emotional question into arithmetic.
| Bankroll | 1% unit | 2% unit | 5% unit |
|---|---|---|---|
| 100 | 1 | 2 | 5 |
| 500 | 5 | 10 | 25 |
| 1,000 | 10 | 20 | 50 |
| 5,000 | 50 | 100 | 250 |
My default recommendation is 1%, and up to 2% if you mostly bet at short-to-medium odds (say under 2.50) and don’t place many bets a week. If you like long shots, parlays or accumulators, go lower, because those lose far more often. The parlay calculator shows how quickly the win probability drops as you add legs, and the parlay guide explains why.
(Unit, yield, stake: if any term here is new, the betting glossary has one-paragraph definitions.)
Units have a second job: they make records comparable. When a tipster says they’re “up 30 units”, that means 30 standard stakes of profit. It also means you should ask how many units they staked to get there, which is a big part of judging a tipster properly.
Flat vs percentage staking
These are the two plans worth your time.
Flat staking means the same cash amount on every bet. Bankroll 1,000, unit 10, every bet is 10 until you deliberately recalculate.
Percentage staking means the same share of your current bankroll on every bet. At 1%, you stake 10 when you have 1,000, 9.90 when you have 990, and 10.50 when you have 1,050.
Here are both plans across the same five bets at odds of 2.00, starting from 1,000:
| Bet | Result | Flat stake | Flat bankroll after | 1% stake | Percentage bankroll after |
|---|---|---|---|---|---|
| 1 | Lose | 10.00 | 990.00 | 10.00 | 990.00 |
| 2 | Lose | 10.00 | 980.00 | 9.90 | 980.10 |
| 3 | Win | 10.00 | 990.00 | 9.80 | 989.90 |
| 4 | Lose | 10.00 | 980.00 | 9.90 | 980.00 |
| 5 | Win | 10.00 | 990.00 | 9.80 | 989.80 |
Over five bets the difference is pocket change. Over a long bad run it isn’t, because percentage staking keeps shrinking the stake and mathematically can’t reach zero. Over a long good run it compounds.
| Flat staking | Percentage staking | |
|---|---|---|
| Effort | Minimal | Recalculate before each bet |
| In a losing run | Stakes stay the same, so losses are linear | Stakes shrink, so the fall slows down |
| In a winning run | No compounding | Compounds |
| Record keeping | Very easy to read | Slightly messier |
| Best for | Beginners, casual bettors | Regular bettors with a log |
A practical middle road: bet flat, and recalculate your unit on the first of each month or whenever your bankroll has moved 20% either way.
Staking plans to avoid
Anything that raises your stake after a loss. The best-known is doubling up after every losing bet until a win “recovers” everything. It looks clever on a napkin. In practice, a run of seven losses turns a stake of 10 into a required stake of 1,280, with 1,270 already lost, all to win 10. Losing runs of seven aren’t rare, as you’ll see below.
“Confidence” staking, where you put 5 units on the ones you really fancy, deserves suspicion too. Unless you’ve tracked your results by confidence level and the high-confidence bets truly do better, you’re just betting more on the matches you’re most emotional about.
What happens in a losing streak
This is the table I’d like every new bettor to see before their first deposit. It shows how much of your bankroll is gone after ten straight losses, by stake size:
| Stake per bet | Flat staking: lost after 10 losses | Percentage staking: lost after 10 losses |
|---|---|---|
| 1% | 10.0% | 9.6% |
| 2% | 20.0% | 18.3% |
| 5% | 50.0% | 40.1% |
| 10% | 100% (bust) | 65.1% |
And ten losses in a row happens more than intuition says. The chance of any particular ten bets all losing is small, but you aren’t placing ten bets. You’re placing hundreds, and every one of them is the possible start of a streak. The table below is calculated for 500 bets, assuming every bet is independent and your win rate never changes, which real betting only approximates. The middle figure is the longest losing run you’d see in a typical set of 500 bets; the last column is a run that still turns up at least one time in ten:
| Your true win rate | Typical odds range | Typical longest losing run in 500 bets | Reached at least one time in ten |
|---|---|---|---|
| 55% | around 1.80–1.90 | 7 | 9 |
| 50% | around 2.00 | 8 | 11 |
| 40% | around 2.50 | 11 | 14 |
| 30% | around 3.30 | 15 | 20 |
| 20% | around 5.00 | 22 | 30 |
These come from a simplified model, not a promise, but the message is solid. Even a genuinely good bettor winning 55% of the time should expect to lose seven in a row somewhere in 500 bets, and has better than a one-in-three chance of losing eight. At a 50% win rate, a ten-bet losing run turns up in about one set of 500 bets in five. With 1% stakes that’s an irritating week. With 10% stakes it’s the end. This is also why a tipster’s hit rate on its own tells you little: the win rate and the odds always come as a pair.
The Kelly criterion, in brief
Flat and percentage staking treat every bet the same. The Kelly criterion doesn’t: it stakes more when your edge is bigger. The formula for the fraction of your bankroll to stake is:
Kelly fraction = (b × p − q) ÷ b
where b is the decimal odds minus 1 (your profit per unit staked), p is your estimated probability of winning, and q is 1 − p.
| Odds | Your win probability | Calculation | Full Kelly | Half Kelly | Quarter Kelly |
|---|---|---|---|---|---|
| 2.00 | 55% | (1 × 0.55 − 0.45) ÷ 1 | 10.0% | 5.0% | 2.5% |
| 2.10 | 50% | (1.1 × 0.50 − 0.50) ÷ 1.1 | 4.5% | 2.3% | 1.1% |
| 3.00 | 36% | (2 × 0.36 − 0.64) ÷ 2 | 4.0% | 2.0% | 1.0% |
| 5.00 | 22% | (4 × 0.22 − 0.78) ÷ 4 | 2.5% | 1.3% | 0.6% |
| 1.50 | 70% | (0.5 × 0.70 − 0.30) ÷ 0.5 | 10.0% | 5.0% | 2.5% |
| 2.00 | 48% | (1 × 0.48 − 0.52) ÷ 1 | −4.0% | no bet | no bet |
Two things jump out. First, when there’s no edge (the last row), Kelly says don’t bet. That’s the most useful output of the formula. Kelly only makes sense alongside value betting, because without positive expected value the answer is always zero.
Second, full Kelly stakes are big. Ten percent of your bankroll on one bet is a wild ride, and it’s only “optimal” if your 55% is truly 55%. If the real figure is 51%, full Kelly has you overbetting badly, and overbetting is what kills bankrolls. That’s why nearly everyone who uses Kelly in practice uses a fraction. Quarter Kelly lands, not by coincidence, close to the 1–2% unit I recommended above: between 0.6% and 2.5% in these examples.
My take: if you can’t put a defensible probability on your bets, skip Kelly and stake flat. A precise formula fed with a guess is still a guess, just with decimals.
A bankroll routine you can actually stick to
- Set the bankroll. Money you can lose. Write the number down.
- Set the unit. 1% by default.
- Set limits at the bookmaker. A deposit limit that matches your bankroll, and a session reminder if one is offered.
- Get the best price. A better price is free bankroll. Compare a few bookmakers and use the odds converter if the formats differ.
- Log every bet. Date, market, odds, stake, result. A spreadsheet is fine.
- Recalculate monthly. New bankroll, new unit. Down as well as up.
- Have a stop rule. For example: down 30% from your starting bankroll means a two-week break and a look at your log before betting again.
- Never top up mid-slump. A reload during a losing run is chasing with extra steps.
What bankroll management can’t do
It can’t make losing bets profitable. If your bets have negative expected value, which is true of most bets because of the bookmaker’s margin, a perfect staking plan just means you lose slowly and predictably rather than suddenly. That’s still worth a lot. Slow, predictable losses are what an entertainment budget looks like. Sudden ones are how people get hurt.
It also can’t fix tilt, the urge to win it all back tonight. The rules above only work if you follow them when you least want to. If you notice you can’t, that isn’t a staking problem, and it’s worth taking seriously.
Betting should cost you no more than any other hobby; if it’s starting to cost more, the responsible gambling page has limit tools, self-exclusion options and help lines.
Frequently asked questions
How much of my bankroll should I bet per game?
For most people, 1% to 2% of the bankroll per bet is the sensible range, and 1% if you bet on longer odds or place many bets a week. At that size, a run of ten straight losses costs you 10–20% of your bankroll, which hurts but leaves you in the game. At 10% per bet, the same run can finish you.
What is a unit in betting?
A unit is your standard stake, expressed as a slice of your bankroll rather than a cash amount. If your bankroll is 1,000 and your unit is 1%, one unit is 10. Units let people with different budgets compare results: a tipster who is 'up 12 units' made 12 standard stakes of profit, whatever a stake means for you.
Is flat betting better than percentage staking?
Neither is better on paper. Flat staking (the same amount every bet) is simpler and makes your records easy to read. Percentage staking (the same share of your current bankroll) shrinks stakes during bad runs and grows them in good ones, so it's slightly safer and compounds. For beginners, flat stakes with a recalculation every month or so gives most of both benefits.
What is the Kelly criterion in betting?
It's a formula that tells you what share of your bankroll to stake based on your edge: (b × p − q) ÷ b, where b is the decimal odds minus one, p your win probability and q the chance of losing. It maximises long-run growth if your probability is right. Because it rarely is, most bettors use half or quarter Kelly.
Should I increase my stakes after a loss?
No. Raising stakes to win back losses is called chasing, and systems built on it, such as doubling after each loss, fail because losing streaks are longer than people expect and both your bankroll and the bookmaker's stake limits are finite. Keep the stake the same, or let it shrink with your bankroll.
How big should a betting bankroll be?
There's no correct number. It should be an amount you could lose completely without it affecting your bills, your savings or your mood for more than an evening. Work backwards from there: if 50 is what you can spare this month, your unit is around 0.50 to 1. If that feels too small to bother with, that tells you something useful.