Bankroll management: a simple framework for staying in control
A practical, plain-language framework for deciding how much money to set aside for betting, how to size wagers, what variance does to even a good bettor, and why limits matter more than any staking formula.
This article is educational and describes how betting markets work. It is not advice, a prediction or an encouragement to bet. Sports betting is legal only in some places and only for adults, and over time the odds favor the house. If gambling is a problem for you or someone you know, call 1-800-GAMBLER in the United States, or look for the help line in your own country.
Ask people who bet on basketball how they decide how much to wager, and most will describe a feeling. This game feels strong, so a bigger stake. Last night went badly, so a bit more tonight to get back to even. The account is healthy, so why not another bet. Few will describe a rule. That gap between feeling and rule is where most of the damage in betting is done, and it is the reason that bankroll management exists as a topic.
Bankroll management is a simple set of habits about money rather than about basketball. It does not tell you what to bet on and it cannot make a losing approach win. What it can do is limit how much a bad run can cost, keep wagers from growing silently, and separate money meant for entertainment from money meant for living. This article lays out a framework in plain language, shows the arithmetic of variance that makes the framework necessary, and explains the one famous staking formula in the field and why it is easy to misuse. It is educational. It is not a system for winning.
What a bankroll is, and what it is not
A bankroll is a fixed amount of money you set aside for betting and nothing else. It is the maximum you are prepared to lose, a budget in the same sense as a night out. Money for rent, bills, savings, debt payments and emergencies is not part of it. The bankroll should be an amount that you could lose entirely without it changing your life, because for most people who bet, losing is the long-term average, as explained in the vig and the hold.
Three rules follow immediately.
- Decide the number before you start. Not after a win, not after a loss. A number chosen in the heat of the moment tends to rise.
- Keep it separate. A dedicated account or a fixed deposit makes the limit physical. Mixing betting money with everyday money makes it easy to lose track.
- Treat it as spent. Think of the bankroll as the cost of the entertainment, like a ticket price. If it returns something, that is a bonus. If it does not, you have paid for the evening.
The last point is the heart of the matter. Betting framed as an investment tends to produce bad decisions, such as adding money to recover a loss. Betting framed as a purchase with a ceiling tends to produce better ones.
Why you need a framework: the arithmetic of variance
The best reason for a framework is that results are noisy. Here is a worked example that surprises most people.
Imagine a bettor who really does have an edge: a true 54 percent win rate on bets priced at −110. That is a very good record, far better than most can sustain, and it implies a profit of about 3.1 percent on the amount staked. Let us call one unit the stake on each bet.
- Each bet wins about 0.909 units or loses 1 unit.
- The average result per bet is 0.54 × 0.909 − 0.46 × 1 = +0.031 units.
- The standard deviation per bet is about 0.95 units.
Over 100 bets, the expected profit is about 3.1 units, and the standard deviation of the total is about 0.95 × √100 = 9.5 units. That is three times the expected profit. The chance of finishing behind after 100 bets, even for this skilled bettor, is about 37 percent. After 1,000 bets, the expected profit grows to about 31 units and the standard deviation to about 30, and the chance of being behind drops to roughly 15 percent. Even after a thousand bets, a bettor with a real, substantial edge has a meaningful chance of being down.
Two lessons follow. First, a short run tells you almost nothing about whether an approach is good. Winning for a month is not proof; losing for a month is not proof. Second, a swing of ten or twenty units in either direction is normal for a bettor of any ability. If your stake is too large relative to your bankroll, a perfectly ordinary bad stretch can end your betting altogether. Management is about surviving ordinary bad luck.
And for a bettor with no edge, which is the typical case, there is no skill to protect. The expected loss is the margin times the amount staked: at −110, about 4.5 percent of everything wagered. Bet 10,000 in total over a season and the expected cost is about 450. The most effective single control on that cost is the total amount staked, not the size of any one stake.
A simple framework
Here is a framework built from habits that are widely used and easy to follow.
Step 1: set a budget you can afford to lose
Choose an amount for a month or a season. Consider what you would otherwise spend on entertainment, and be conservative. Write it down. If you are not sure whether you can afford it, reduce it.
Step 2: define a unit
A unit is your standard stake. Many bettors define a unit as a small fraction of the bankroll, commonly between 1 and 2 percent, although it can be smaller. With a bankroll of 500, a 1 percent unit is 5. The point of the small fraction is survival: even a streak of twenty straight losses, which will happen sooner or later if you bet often enough, costs 20 units and not the whole bankroll.
Look at the arithmetic of a losing run. With a 52 percent chance to win each bet, the probability of losing ten in a row is 0.48 to the tenth power, which is about 0.065 percent, or one in 1,500, for a single sequence. But across hundreds or thousands of bets, there are many possible starting points, and a streak of ten or more losses becomes quite plausible over a long period. Small units absorb that. Large units do not.
Step 3: keep stakes flat
Flat betting means staking the same number of units on every wager. It keeps the process simple and removes the temptation to increase stakes when you feel confident or after a loss. Both of those moods are unreliable guides. Confidence has little relationship with whether a bet is actually good, and increasing stakes after a loss is the most common form of chasing.
If you vary stakes at all, do so by rule, not by feeling, and keep the range narrow, for example between half a unit and one and a half units.
Step 4: cap the exposure
Set a limit on the number of bets or the total at risk on a given day or week. A cap prevents a rush of wagers on a busy night from taking more of the bankroll than you intended. It also stops live betting from turning into a stream of small decisions, a pattern discussed in our article on live betting. Complex tickets deserve extra caution. Parlays lose far more often than single bets, as explained in our piece on parlays and teasers, so many people give them a much smaller fraction of the bankroll than their straight bets.
Step 5: decide the stopping points in advance
Choose two lines before you begin. One is a loss limit: the point at which you stop for the day, the week, or the month. The other is a time limit: how long you will spend betting. When a limit is reached, stop. Do not renegotiate with yourself.
A win limit is optional, but some people find it useful: a point at which you stop and withdraw, so winnings do not simply flow back into the next round of bets.
Step 6: keep a record
Write down every bet: date, market, price, stake and result. Records show what is really happening, not what you remember. Memory favors the wins and the dramatic near misses. A record also lets you compare your prices with closing prices, which is the most reliable available test of whether you are beating the market, as discussed in closing line value. Most people find that records are humbling, and that is part of their value.
Step 7: never chase
If a night goes badly, the plan does not change. You do not increase the next stake, add a late bet to get back to even, or raise the bankroll. Chasing losses is the clearest warning sign described in our guide to responsible gambling. If you feel the urge, that is a signal to stop for the day.
A worked example of the framework
Here is how the steps fit together for someone who decides to bet small amounts on basketball for a season, using made-up numbers.
- Budget. The person sets aside 400 for the season, an amount they would be comfortable spending on entertainment. It sits in a separate account and is never topped up.
- Unit. One unit is 1 percent of the budget, so 4. Most wagers are one unit. Parlays, which cost more per dollar staked, get half a unit.
- Cap. No more than five units at risk on any single day, which means that even a bad night costs at most 20.
- Stop rule. If the account falls to half its starting value, the person stops betting for a month and reviews the records before deciding whether to continue at all.
- Record. Every bet goes into a simple log with the price taken and the closing price.
Now watch what the rules do. A cold streak of fifteen straight losing units costs 60, or 15 percent of the budget. It is unpleasant but survivable, and it does not trigger any desperate change. Without a rule, the same cold streak might prompt a doubling of stakes, which turns a loss of 60 into a loss of 120 or more in the same number of bets. The framework does not improve the quality of the picks. It changes the cost of being wrong.
Notice, too, what the example leaves out. There is no step that says "increase the unit after a win" and no step that says "reload the account." Those omissions are deliberate. Most of the harm in betting comes not from the first stake but from the decision to go beyond the plan.
The Kelly criterion, and why to be careful with it
Any article on staking will eventually meet the Kelly criterion, a formula for the fraction of a bankroll that maximizes long-run growth when you know your edge precisely. The standard form is:
f = (b × p − q) ÷ b
where f is the fraction of the bankroll to stake, b is the net payout per unit staked (0.909 at −110), p is the probability of winning, and q is 1 − p.
For the skilled bettor in our earlier example, with p = 0.54, the calculation is (0.909 × 0.54 − 0.46) ÷ 0.909, or about 3.4 percent of the bankroll per bet. At exactly the break-even probability of 52.4 percent, the formula gives zero. Below it, the formula gives a negative number, which is an instruction not to bet.
Kelly is elegant, but there are two serious problems.
- It depends on knowing p. The formula takes your estimate of the win probability as exact. In real life, an estimate is uncertain, and estimates made by bettors tend to be too optimistic. If you believe the chance is 54 percent and it is really 52 percent, the formula's recommended stake becomes a wager with negative expected value. Overestimating the edge leads to overbetting, which hurts more than underbetting.
- It produces large swings. Even when the estimate is right, full Kelly produces large drawdowns. Many practitioners therefore use a fraction, such as half Kelly or less, to smooth the ride.
The honest summary is that Kelly is a useful concept for understanding the trade-off between risk and growth, and a risky formula for beginners to apply. For anyone who is not certain of an edge, and nearly nobody can be, a small flat unit is more robust. The idea behind Kelly, that stakes should shrink as the edge shrinks, is the part worth keeping, and it ties directly to expected value.
Common mistakes
- No fixed bankroll. Betting from a general account makes the true cost invisible.
- Raising stakes after wins. A winning streak is not evidence of skill. It is often variance, and larger stakes magnify the reversal.
- Raising stakes after losses. This is the most dangerous habit. It turns a manageable loss into an unmanageable one.
- Treating bonuses as free money. Promotional credits come with conditions. They can encourage more betting than planned. See sportsbook limits, boosts and promotions.
- Overcounting skill. Believing a short run proves an edge leads to overconfidence. The 100-bet example above shows how little a short run says.
- Ignoring time and mood. Betting when tired, upset or intoxicated is a reliable way to abandon a plan. Many people set a rule not to bet in those states at all.
- Spreading across accounts without tracking. Having several accounts is useful for comparing prices, as in shopping for lines, but it can also hide total spending. Track the combined figure.
Tools that help
Many sportsbooks provide controls that do part of this work: deposit limits, loss limits, time reminders, cooling-off periods and self-exclusion. Using them is not an admission of a problem. It is the same logic as a budget, implemented where the temptation occurs. Setting them when you are calm is much easier than trying to hold to a limit in the middle of a bad night.
If you bet and feel your control is slipping, talk to someone. In the United States the National Problem Gambling Helpline is 1-800-GAMBLER, and help lines in other countries are easy to find. The page on responsible gambling describes the signs to watch for.
What to take away
Bankroll management is about limits. Decide what you can afford to lose, treat it as spent, keep stakes small and flat, set loss and time limits before you start, record what you do, and never chase. The mathematics of variance explains why: even a genuinely skilled bettor can be behind after a hundred bets, and a bettor without an edge pays the house margin on everything staked. No staking plan changes the sign of the expected value, but a good one keeps a bad run from becoming a disaster. If you are looking at prices on the odds page purely out of interest in the game, none of this applies. If you are wagering, it applies to every bet.
If betting stops being fun or starts to cost more than you can afford, stop and talk to someone. In the United States the National Problem Gambling Helpline is 1-800-GAMBLER, and most other countries have a similar service.