Meistro said:
You see, the Kelly Criterion is a recipe for bankroll growth for a series of even money wagers.
Where did you come up with that?
Meistro said:
By betting full kelly, that is to say by betting your advantage, you will experience the most rapid bankroll growth possible
Full Kelly is the bet which maximizes the logarithm of your expected bankroll after the bet is resolved. It is not simply betting your full advantage. Advantage divided by variance is a decent approximation of this. For blackjack, since the variance is approximately 3.1 for typical games, multiplying by 76% is also a reasonable approximation.
Meistro said:
But because blackjack is not a series of even money wagers, it is important to bet less than your advantage, otherwise you will be betting more than Kelly
Games in which you do not change your bet also have variance. There are some +EV video poker games. The optimal Kelly bet for these would also consider both the EV and the variance. Hole carding 3 card poker and other carnival games is +EV and has variance and thus a correct Kelly bet based on both
Meistro said:
Risk of ruin (ROR) is something that can be simulated.
I know some may consider this pedantic, but I do not. By the definition in BJA3, which is the definition we use when discussing optimal betting and Kelly, RoR is the probability of going broke if you play forever with your bankroll, never resizing your bets. It is not the commonly misinterpreted probability of going broke vs doubling your bankroll. The latter also exists, is different and can be simulated. The former cannot be directly simulated without some sort of stop heuristic for the cycles where the player never goes broke.
It is more correct to say that RoR can be
calculated (using the formula I gave above) from data (EV, Variance) collected by a simulator.