[jbrennen]

There's a concept called the Kelly Criterion for achieving optimum long term growth in a betting game where you have a quantifiable edge, and where you can choose your bet size at will.

Essentially, a modified version of the Kelly Criterion, applied to poker, indicates that for a given winrate and standard deviation (in BB/100), you achieve optimal growth when your bankroll is equal to the square of your standard deviation divided by your winrate.

To use the numbers from earlier in the thread, if you have a winrate of 1 BB/100 and a standard deviation of 15 BB/100, you will achieve optimum growth if you're playing with a bankroll of 225 BBs (15 times 15 divided by 1).

In practice, this means that you'd take your bankroll, divide it by 225, and play a stake level which comes close to that number. If your bankroll is $600, for instance, you'd come up with a big bet size of $2.67; the nearest game to that would be $1/$2, so that's where you should be playing.


One obvious thing about using the Kelly Criterion -- you need to move up or down in stakes when the formula says that you should, if you intend to play at the optimum level. If you're that guy with the +1 BB/100 winrate, 15 BB/100 standard deviation, and your bankroll falls below $337.50, now that your optimal big bet size has fallen below $1.50, you need to move down to the $0.50/$1 game.

Note that the risk of ruin with this method is theoretically zero, because you'd keep playing smaller and smaller stakes as you lose money. In the real world, there's a minimum stake level where you could go broke, of course. But if you start with $600, if you keep moving down in stakes when appropriate, and if you still go broke, you need to re-examine the idea that you actually are a winning player...