Comp Sims VS Hand-Dealt Shoes

Until someone hand deals a million hands there's no way to know for sure that comp sims are an accurate indicator of the way physical cards will behave.

Maybe so . . . maybe not . . . but the bottom line is that it's an assumption until proven.

Series of random numbers can be substituted for one another with the same results. I don't know who did the proof of that but I strongly suspect it has been done, because in the definition of random any segment of any random series is indistinguishable from another.

Take a random series of digits 0 through 9...

The digits 0, 1, 2, 3,... 7, 8, 9 will all appear with the same probability

The pairs 01, 02, 03,... 97, 98, 99 all will appear with he same probability

001, 002, 003,... 997, 998, 999........ and so on indefinitely

That said, the sim results will match the hand dealt results insofar as each is random. We know the physical cards are not random but they are close enough that we don't see any gross effect from it. The decent sim programs use a RNG that's as close as you can get to random and still run on a PC.

The CV software lets you define the shuffle used in the simulations. You can program it to shuffle just like they do at your local casino. The long-run results of the "human" shuffle were identical to the results of the RNG shuffle. Case closed. Because of the random cut, the "human" distribution will be identical to a random distribution.

-Sonny-

Well I daresay it's not "case closed" until someone has actually taken the time hand-deal a million hands and offer proof. Call me stubborn.

Also Sonny, I haven't been around here in a while and I will say that in your latest posts I've run across I've noticed a combativeness and irritation in your tone that seems to be a recent development.

No, you’re not being stubborn. Stubborn would be refusing to change your mind despite the proof. You’re exactly the opposite: not willing to give up until you see definitive proof. That’s what I call being smart.

Yeah, that’s a recent development. There has been a lot of activity on the Voodoo forum lately. Let’s just say that some of it has really been trying my patience. Actually, I’m glad that you mentioned that. I apologize to everybody if my frustration has overflowed onto the other forums as well. I just need to take a deep breath…and a vacation from progression systems.

-Sonny-

Call me a skeptic. I'm sure it has something to do with my religious past and my previous tendency to buy into all sorts of beliefs or ideas. Now I ONLY believe much of anything when an unbreakable proof is offered. Otherwise it just drifts in the sea of possibilities.

I am also well aware of the fact that what SEEMS to be true is not always true, and there have been many things in the world that have been declared impossible that were later show to be very impossible indeed. Telephones . . . televisions . . . automobiles . . . airplanes . . . space ships . . . microwaves . . . the Internet . . . all unthinkable in the comparatively recent past.



Well I do like the voodoo forum. Call me a fool but lately I have been quite enchanted by Jay Moore's ideas put forth in his book Most Powerful Blackjack Manual.

While I do believe the title is misleading and he has some questionable ideas when it comes to basic strategy, his betting method is worth a look. It's definitely different. Instead of a straight progression he calculated all 64 possibilities in regard to wins and losses that can occur within 6 consecutive hands of play and developed a betting method in which 54 of those possibilities will end with a profit.

And according to the author (who makes himself available via e-mail): "The 64 possibilities adds up to a positive result, if you consider the first row is open to a long winning streak."

It's probably also worth mentioning that he is A) a mathematician, and B) a former counter who claims that he was successful as a counter but not really in comparison to the amount of effort he had to put into it. He abandoned counting for his betting method, which I hesitate to call a progression.

And I really got off track there. I guess what I really wanted to say was that what is considered voodoo today will be reformulated tomorrow to become accepted truth in the future. So I think it's always good to keep a skeptical, but open, mind.

Almost certainly, the results would not be identical anyway. How close must they be to make you believe it or not?

I don't really know but I would guess one could have a very high level of confidence, or not, after many fewer hands than 1000000.

Haven't read it, don't know the rules of the game he's talking about but it sounds like he's probably ignoring ties, He's probably only counting net wins or net lossess rather than total units won or lost and probably ignoring splits and doubles as a start.

Does 6 hands of play mean 6 dealer upcards?

And, apparently, I'm guessing he's assuming a win or loss as equally likely.

Neither here nor there I guess. And it is good to keep an open mind.

But right away I'm skeptical lol.

Yes, and that will produce a positive expectation if all 64 outcomes have an equal probability of occurrence. Unfortunately, that is not the case with BJ (or any other game where the house has the edge). Of those 64 outcomes, the ones with losses will happen more often than the ones with wins. All 64 outcomes are not equally weighted. Even though most of the outcomes involve a win, after you factor in the frequencies of occurrence you will find the house edge is still there. Just like with any progression system, it might win more often but its overall results will still be negative. That’s not always a bad thing, but it can be a little misleading sometimes.

ANYWAY…back to the shuffling topic.

There have been numerous studies as to the effectiveness of human shuffles and the simulation of human shuffles by computers. Some of the well-known studies can be found on Snyder’s website:

http://www.blackjackforumonline.com/content/manvscom.htm
http://www.blackjackforumonline.com/content/random.htm

He also references the work of John Gwynn and Persi Diaconis and D. Aldous. There has also been work published by Richard Hannum, Steve Forte, Richard Epstein, and probably many others that evade my memory. Although everybody agrees that individual casino shuffles are nowhere near random, there is no indication that their overall distributions are not normal or that computers cannot accurately simulate their results. According to Steve Forte (who has been analyzing both human shuffles and CSM shuffles for many years and helped design a CSM machine):

“The contention that computers can’t be programmed to emulate the human deal, card pickup, and shuffle is untrue. On the contrary, the process of dealing and picking up the hands is mechanical, definable, and easy to program. And when it comes to shuffling, computers do a wonderful job of emulating the process, provided the programmer understands that process…Emulating the human shuffle is relatively simple, and any level of detail or sophistication is attainable, if desired.” (p.219-220)

-Sonny-

He ignores pushes and counts doubles and splits as two hands.

One thing I can definitely say is that regardless of the systems intrinsic value, it took some intelligence to formulate.

I think it's worth a look if you are at all interested ideas regarding betting that have nothing to do with counting.

I hear you. Truth be told, if I were a betting man (ha ha) I would probably put my money on Yes, a computer can simulate a hand-dealt shuffle.

But I still don't KNOW it. I may suspect it, I may consider it likely, but it's still just an assumption which leaves nagging doubts.

One person is never going to hand-deal a million hands but there is a way it could be done. If you could get many people to shuffle and hand deal 1,000 hands--with everyone using the exact same shuffle method and playing the hands according to perfect BS--you could then collect the results and you would have a million hand-dealt hands.

You would just have to have people that you could trust to do it properly and not f**k it up.

Statistical significance of sims

I have an article on my website about how many possible equivalent shuffles there are in the game of blackjack.
A billion round sim doesn't even come close to the amount of possible shuffles and each shuffle possibility can account for many rounds.

I'm not saying that a billion round sim isn't significant, but am just looking for a little perspective. I've never taken a statistics class, but I'm sure it's a valid science. Election results are predicted on a relatively small data sampling. I've also heard that a good statistician can make data say what he wants. How many times have you heard that statistics have shown a certain food is detrimental and then maybe a year later hear that statistics have shown that same food to be benefical?

Just some food for thought:grin:
k_c