Originally Posted by
Wrecktangle
Look, Full Kelly is mathematically provable as the optimal way to manage your bankroll. Dr Kelly showed this in the 50s while working for Ma Bell.
Where everyone fails while using it is: understanding the conditional win% even within one sport. This can vary according to part of the season, injuries, management and coaching environment, team rest, weather, etc. If you are using a win% across multiple sports (totals and sides in one sport, is really two sports), you are making a mistake.
I know of no one who stays on top of all this and then gets into the Bayesian aspect of it. Until you do this, you will not know your true win %, and you are setting yourself up for an eventual BR crash. It is perhaps not as bad as the Progressive betting systems, but it can get close.
if you do not keep extremely accurate records, you cannot use Kelly.
If you do not understand your conditionals, you cannot use Kelly.
If you do not understand Bayesian Statistics, you cannot use Kelly.
These three conditional statements rule out 99.99% of all bettors.
So given this, almost everyone is better off using flat betting, or a very small fractional Kelly. Otherwise you risk wiping out the bulk of your BR before the end of the season.