computer perfect play for online live dealer games

Meistro

Well-Known Member
#1
It occurs to me that computer perfect play might be a way to make money at online live dealer games. The program necessary for such a game would be pretty simple. All the program would need to do is start with a whole shoe, and then enter in the rank of each card that is dealt and subtract that rank from the shoe array. Then the computer would need to simulate the various playing possibilities given the remaining deck composition and should be able to make more intelligent hit / play decisions. The computer could also simulate the present deck composition and more sharply determine the player's advantage or disadvantage.

Am I right in thinking that computer perfect play would give the player a nice edge, even with not so great penetration? If an over all edge of action of say 2% could be obtained, then you wouldn't even need to risk that much money in terms of trip roll on hand in case the casino decided to ban you and confiscate your funds. Or am I over estimating the edge from computer perfect play? It seems to me that a computer should be able to much more accurately calculate changes in advantage based on deck composition through simulation than any crude counting system could hope to accomplish, especially when you count on the ability to make much more accurate hit / stand / double decisions. Or is poor penetration enough to stifle even the nimblest mechanical mind?
 

Meistro

Well-Known Member
#3
I presume, based on my reading and interpretation of said discussion, that the authors of it would answer 'no' to my question of whether a sizeable advantage can be gained through computer perfect play in the 4/8 games that proliferate in online live dealer games today. Still, I would be interested to see the results of a simulation that tests this hypothesis. I'm tempted to try to program such a device myself, although I haven't toyed with computer programming since I was just a boy.
 

gronbog

Well-Known Member
#5
You don't need to run a simulation against the remaining cards in the shoe. It can be done by combinatorial analysis. You don't need to write that program either. Eric makes the code for his CA software available online.

While Don is right that there would not be an advantage for flat betting using this information, we do believe that using the information perfectly for both betting and playing would represent the limit on what could be achieved via card counting alone.

Finally, the flat betting implementation of this was done decades ago for live play in real casinos using wearable computers.
 
#6
gronbog said:
It can be done by combinatorial analysis. You don't need to write that program either. Eric makes the code for his CA software available online.

While Don is right that there would not be an advantage for flat betting using this information, we do believe that using the information perfectly for both betting and playing would represent the limit on what could be achieved via card counting alone.
So for a 50% pene on 6D slow game...what could it be?

.
 

gronbog

Well-Known Member
#7
Unfortunately, Eric's work in this area seems to have stalled. Although each decision can be computed by CA, a sim is still needed to obtain the resulting SCORE. Last I heard, the thinking was that the sim would be prohibitively slow using today's technology.
 
Top