The Kelly Criterion is a famous formula for deciding how much of your bankroll to wager to maximize long-term growth. Professional gamblers and investors use it when they have an edge. Understanding what it says about casino games is one of the most clarifying things you can learn as a player, and the conclusion is blunt but useful.
What Kelly actually says
Kelly sizes your bet based on your edge and the payout odds. The critical detail: when your expected edge is negative, as it is on every casino game by design, the Kelly-optimal bet size is zero. Mathematically, the way to maximize your bankroll on a negative-edge game is to not bet at all.
So why play?
Because you are playing for entertainment, not as an investment, and that is a completely valid reason. The honest framing is this: you are paying the house edge as the price of playing, the same way you pay for a movie ticket. Kelly does not tell you to never have fun. It tells you to be clear-eyed that these games are a cost, not a strategy for growing money.
The practical takeaway: fractional, tiny bets
Since full Kelly says zero, the sensible real-world approach is the opposite of what most players do. Instead of chasing losses with bigger bets, you keep every wager to a small, fixed fraction of your bankroll, typically 1 to 2%. This does the one thing you can control: it maximizes how long you get to play on a given bankroll and softens the swings, so a bad run is a slow afternoon rather than a wipeout.
What this means for systems
It also explains why aggressive progression systems like Martingale are so dangerous. Doubling after losses is the exact opposite of small fractional betting, and on a negative-edge game it accelerates ruin. See why Martingale fails on Stake and our bankroll management guide for the safer approach.
Bet the smart way, not the loud way
Our sheets turn fractional bet-sizing and stop rules into a simple system you can follow.
For entertainment only. 18+. No strategy can guarantee a profit or overcome the house edge. Gamble responsibly.