What a bankroll is
In probability terms, a bankroll is simply the amount of money set aside for a series of bets or plays. Bankroll maths is the arithmetic of how that amount interacts with the size of each stake and the swings of the game. It is a way of describing risk. It is not a way of removing it, and it cannot turn a game with a negative expected result into a profitable one.
Measuring in units
The first idea is to stop thinking in currency and start thinking in units, where one unit is the size of a single stake. A budget of 100 units with a one-unit stake can absorb 100 straight losses. The same money with a ten-unit stake is only 10 units deep.
Nothing about the odds changes. What changes is how many plays the budget can survive, and how large the swings look relative to it.
Expected loss scales with stake size
Where a game has a house edge, the expected result per play is the stake multiplied by that edge. As an illustration, take an even-money bet on a single-zero roulette wheel. The chance of winning is 18/37 and the expected result is -1/37 of the stake, about -2.7%.
- 100 bets of 1 unit: expected result is about -2.7 units.
- 100 bets of 10 units: expected result is about -27 units.
Ten times the stake means ten times the expected loss. The edge applies to the total amount wagered, which is why the long-run cost of a game grows with how much is staked, not just how often you win.
Swings scale differently
The swings do not scale the same way as the number of plays. For independent even-money bets of one unit, the standard deviation of a single bet is about 1 unit, and the standard deviation of the total after n bets is about the square root of n, in units.
- After 100 one-unit bets: about 10 units.
- After 100 ten-unit bets: about 100 units.
So with a stake ten times larger, the typical swing is ten times larger too. Compare each with the same 100-unit budget. A one-unit stake has a typical swing of around a tenth of the budget over 100 bets. A ten-unit stake has a typical swing about equal to the whole budget. This is a loose description of scale rather than a forecast, but it shows why the same game can feel gentle at one stake and punishing at another.
Why long runs are not safer for the player
It might seem that longer play smooths things out. It smooths the average result per bet, but the expected total loss keeps growing with every extra play. With 10,000 one-unit bets the expected result is about -270 units (10,000 / 37), while the standard deviation is about 100 units. The expected loss has now overtaken a typical swing. That is the arithmetic behind the long-run advantage of the house, and it is why a longer session makes the average outcome worse rather than better.
Where the idea is used
The same vocabulary appears in many fields: investors talk about position sizing, and poker players talk about the number of buy-ins they hold. It is also related to the Kelly criterion, a theoretical rule for stake size when a bet genuinely has a positive edge. In casino games the edge is negative, so there is no stake size that makes the long-run expectation positive.
Bankroll maths also does not rescue common mistakes. Raising stakes after a loss to recover it, chasing losses, simply increases the amount at risk and the swings together. It does not change the odds of the next play. Likewise, a series of wins does not mean a run is due to end, nor that it will continue; each play is independent.
What the numbers are good for
Used honestly, bankroll maths helps with three things:
- Understanding why a budget can disappear far faster at higher stakes than expected.
- Recognising that the expected cost of play is proportional to the total amount staked.
- Thinking in advance about how much one can afford to lose, which is the idea behind spending limits.
It cannot predict the next result, guarantee a session outcome or produce a profit from a negative-expectation game. Results are uncertain, and past outcomes do not change future odds.
If gambling is causing financial or emotional strain, or if you are tempted to borrow to play, please reach out to a local support service. Only wager money you can afford to lose, and treat any loss as a cost rather than something to win back.