Bankroll Management: Units, Flat Betting and Ruin

A bankroll is a defined pot of money set aside for betting, kept separate from everything else, and sized so that losing all of it changes nothing about your life. Rent is not bankroll. Money you would have to explain is not bankroll.

The separation is mechanics, not moralism. The entire point of a bankroll is that it lets you keep staking the same amount through a losing run — and money you cannot afford to lose cannot do that. It forces you to stop at the worst possible moment, or, worse, to bet bigger trying to get back to level.

A unit is one standard bet, expressed as a percentage of the bankroll rather than a dollar figure.

unit = bankroll × unit size

  $2,000 bankroll at 1%  →  $20 unit
  $2,000 bankroll at 2%  →  $40 unit
  $500 bankroll at 1%    →  $5 unit

Units exist because they make records comparable and staking automatic. “Up 14 units” means something to anyone at any bankroll size; “up $280” means nothing without knowing what was risked to get there. It is why every honest public record — including this site’s — is reported in units.

One percent is the usual recommendation, and the reason is the arithmetic of losing runs rather than caution for its own sake. The bankroll calculator turns a bankroll and a unit size into the figure to actually stake.

Flat betting means every bet is one unit. Not two units on the ones you like and half a unit on the ones you are unsure about — one unit, every time.

The case for it is uncomfortable and simple: most bettors cannot rank their own bets. Varying stakes by confidence helps only if confidence correlates with being right, and for most people it correlates instead with how vivid the reasoning felt. Staking more on those turns a small edge into a smaller one, because the extra money goes disproportionately to the bets where the estimate is most inflated.

Flat betting also keeps your own record readable. If every bet is the same size, your win rate and your profit describe the same thing. Once stakes vary, a good record can be one lucky large bet and a bad one can be a single mistake — and no amount of staring at the total will separate them.

Varying stakes is defensible when the edge is measured rather than felt, which is what the Kelly criterion formalises. It scales the stake to the size of a known edge, and it is only ever as good as that input.

This is the section that changes how people stake, because the numbers are not intuitive.

Take a bettor at exactly 52.38% at −110 — precisely break-even, so the arithmetic is not being generous. Here is how often a losing streak of a given length shows up at least once.

Five losses in a row is not a slump. Over a hundred bets it is the expected experience, arriving about three times in four. Seven in a row is a coin flip over 250 bets. Ten in a row happens to more than a quarter of bettors over a thousand.

And this is a bettor with no edge and no bad luck — just randomness. Add the ordinary experience of being slightly wrong and the runs get longer. The streak calculator runs these for your own win rate, and the drawdown calculator converts a run into what it costs at your unit size.

At 1% units, ten straight losses costs 10% of the bankroll: unpleasant, entirely survivable. At 5% units it costs half of it — and a bettor now staking 5% of a halved bankroll is in a hole they cannot climb out of at the rate they fell into it.

Risk of ruin is the probability that a losing run takes the bankroll to zero before the edge takes it up. It depends on three things: how large the edge is, how large the bets are relative to the bankroll, and how long you intend to keep going.

The relationship with bet size is the important one, and it is not linear. Doubling the unit size does considerably more than double the risk, because ruin requires a sequence of losses and larger bets shorten the sequence needed to get there. This is why 1% versus 5% is not “five times more aggressive” but a different category of outcome — the risk of ruin calculator shows the curve.

The related question — not “will I go broke” but “how bad does it get on the way” — is what the bankroll simulator answers, running the same edge thousands of times and reporting the spread. The tenth-percentile season is almost always worse than people plan for, and it is the one a bankroll has to survive.

• Set the unit once, then leave it. Recalculate on a schedule — monthly, or every hundred bets — never after a result. Resizing on Sunday because Saturday went badly is chasing with extra steps.
• Never chase. Raising stakes to recover losses is the most reliable way to turn a survivable drawdown into a terminal one. The bets do not know what you lost yesterday.
• Let the unit move with the bankroll. A percentage of a growing bankroll compounds; a percentage of a shrinking one is what stops a bad run from ending things.
• Same size on every bet — especially the obvious one. That is exactly where the confidence premium does its damage.
• Cap the daily exposure. Correlated bets on one day are effectively a single larger bet, and a fifteen-bet Sunday at 1% is a 15% day whatever the tickets say.
• Do not top up mid-run. Adding money to rescue a drawdown replaces the one honest signal you have with a reset button.