flaw with kelly criterion?

[Thunder]

How does Kelly resolve the issue of playing a game where the loss amounts varies?

 

[blackjack avenger]

no flaw in theory

With bj the skewed payouts can be accounted for
We divide advantage by variance of the game.

The problem with Kelly and bj has to do with table minimums and very long run number. Plus the risk of large drawdown.

 

[MangoJ]

It doesn't.

The classical Kelly criterion only applies to a two-way bet of fixed odds with known probabilities.

If your bet has several outcomes (when won or when lost), you can always calculate a generalized Kelly bet by maximizing <log(bankroll + f*X)>.

Kelly bet has no "flaw", it just isn't applicable to your type of bet.

 

[zengrifter]

It doesn't.

The classical Kelly criterion only applies to a two-way bet of fixed odds with known probabilities.

If your bet has several outcomes (when won or when lost), you can always calculate a generalized Kelly bet by maximizing <log(bankroll + f*X)>.

Kelly bet has no "flaw", it just isn't applicable to your type of bet.

As long as he can estimate his EV it IS applicable, no? zg

 

[London Colin]

I find this to be a moderately understandable explanation (assuming it's accurate) -

http://www.bjmath.com/bjmath/kelly/break.htm (Archive copy)

 

[Thunder]

As long as he can estimate his EV it IS applicable, no? zg

Not nearly as accurate though.. You could know your EV, but because the variance is huge, your Kelly bet wouldn't be correct if I'm not mistaken.

 

[blackjack avenger]

Thunder Clarify

Kelly does apply to bj.
At least there is a whole lot written about Kelly & bj.

We can lose spl double hands
Which is a larger loss then our original bet
Yet Kelly applies

Thunder what exactly are u referring to?