Risk of ruin: why a 99% game still empties a bankroll

The edge is not the story

Every honest casino publishes a return figure, and players read it as a promise about their evening. It is not. A 99% return means that across a very large number of bets, the total wagered comes back at ninety-nine cents on the dollar. It describes the long run and says nothing about your run.

Consider two players with the same 100 units and the same 99% game. One bets 1 unit a hundred times. The other bets 50 units twice. Both face the identical edge. The first will almost certainly still be playing at the end; the second has a serious chance of being finished after two rounds. Nothing about the edge distinguishes them.

What distinguishes them is variance — how far individual results scatter around the average. And bankroll survival is governed almost entirely by that scatter relative to the size of your stack.

The two levers

Strip the mathematics down and only two things determine whether you go broke: how large your bets are compared to your bankroll, and how many of them you make.

• Bet size relative to bankroll. This is the dominant term by a wide margin. Halving your bet size does not halve your risk of ruin; it collapses it, because ruin requires a losing streak deep enough to consume the stack and each halving doubles the depth required.
• Number of bets. More bets means the edge gets more chances to express itself. Against a negative edge, playing longer moves you toward the expected loss with increasing certainty — the law of large numbers is working, just not for you.

The uncomfortable implication is that the two levers point in opposite directions for anyone hoping to leave ahead. Betting small protects the bankroll but requires many bets, and many bets is exactly what grinds it down. Betting large shortens the exposure but invites ruin from a normal run of bad luck.

Where variance actually lives

Two games can share a return figure and feel nothing alike. The difference is how the return is delivered: many small payouts, or rare enormous ones.

A high-variance game needs a much deeper bankroll to survive the same number of rounds, because its losing stretches are longer by construction. If most of the return is concentrated in rare events, the periods between those events are where bankrolls die.

None of this makes high-variance games worse. It makes them a different product, and the mistake is bringing a low-variance bankroll to one.

What this actually implies

The practical conclusions are unglamorous, which is why they rarely get repeated:

1. Set the session budget first, then divide it into bets — not the other way round. A bet size chosen from the bankroll is a decision; one chosen from how you feel is a mood.
2. Keep single bets small relative to the stack if you want to still be playing later. The relationship is not linear in your favour, and that asymmetry is the only thing on your side.
3. Match the bet size to the game’s variance, not to its return figure. A capped-jackpot game and a steady-ladder game need different sizing at identical returns.
4. Accept that no sizing rule turns a negative edge positive. Sizing changes how long the money lasts and how wild the ride is. It does not change the direction.

That last point is the one worth carrying out of the article. Variance is the thing you can manage; the edge is the thing you are buying. Managing the first well makes the second affordable and the evening longer, which — for a game — is the honest definition of doing it right.