Quote from: SebastianJu on December 29, 2013, 06:59:27 PM

Quote from: maqifrnswa on December 29, 2013, 03:36:53 PM

Quote from: SebastianJu on December 21, 2013, 08:55:16 PM

I have read wikipedia about kelly criterion but i dont really get how much bigger the risk is when using kelly 2% instead 1%. I mean doesnt it mean for JD that at kelly 1% 1% of the house can be won. I see JD is only at 0.5% what is even below the optimum kelly describes. But even with 2%... its a percent value... the house cant really be beaten this way or am i wrong? It can go down very much but each new bet with max win of the house 2% means the house will be less than before... so the max win will be less then. If it would be a fixed value then of course... at 2% the house could be beaten in 50 wins. Which sounds like it will happen very rare. But its not even a fixed value. Its a percent value. That means you cant win back the losses as fast as you could with fixed values but the same goes with the losses. They will slow down. I dont see yet why 2% should be so much worse than 1%.

read:
http://wizardofodds.com/gambling/kelly-criterion/

(it's great site by the way, in general)

and:
http://compoundingmyinterests.com/compounding-the-blog/2012/10/12/how-did-ed-thorp-win-in-blackjack-and-the-stock-market.html

and
http://www.bjmath.com/bjmath/kelly/kellyfaq.htm
(see Q5)

If edge is 1%, and you wager 2% every time, the house bankroll won't grow. Wager >2%, and it will actually shrink over time.

He claims that at 2% no growth will happen... really didnt thought that and its still hard to believe... maybe i should run some script to test this out. Anyway... this would only apply if someone plays really the max profit. If he plays half max profit then 2% kelly would be 1% kelly in reality...

I cant find out the formula values to get to 1% as ideal kelly value for just-dice. What are the values for the formula?

the formula is (house edge / bet variance)
house edge = 0,01
variance depends on bet type, with standard 2x (0,495 chance to win) bets you have variance = 1

so thats where the 1% as ideal number for max win comes from, now not all bets are made as standard 2x (the lowest variance bets) so depending on what percentage of bets are made with higher variance, the more we should lower the optimal max win number.

TLDR: optimal max win for just dice is not 1% but <1%