trouble with six deck games

I have currently been working on an analysis of six deck games since where I come from in the midwest that's almost all they play. I have played over 20,000 hands each on my computer using hi-opt1 and the same with a straight +/- betting strategy both head to head. Although a total of 40,000 hands is a relatively small number for statistical purposes It would still take me several years to play that many hands in a casino and if I'm on a losing streak for several years I certainly should retire. So in escence the results should be acurate. Anyway some of this may seem strange but, I have found that there is no advantage to counting in a six deck game over perfect basic strategy and solid betting structure. The true count rarely went over +2 and that of course was almost always in the last few deals. The results were within only a few percent of each other in winnings with the slightly higher percentage in the non counted sessions. Speaking of winnings I have also found that not only is winning very difficult in six deck play, it's nearly impossible playing head to head. This would contradict everything I've read about the game, however none of the literature was based on six decks. In head to head play with six decks I experienced many severe losing streaks as much as 17 hands in a row, however my winning streaks at maximum were 8. When I play with a full table my losing streaks rarely exceed 7 or 8 and of course I quit immediately if they do. But interestingly enough my winning streaks with a full table remain the same at 8. Although I have been able to beat the game both ways and amazingly head to head as well, I tend to feel that you have an equal chance wether counting or not and you are better off playing with more people at the table when playing six decks. Anyone have similar results, or am I completely out of my mind? As well I am looking for literature that has statistics on six deck play.

Today, high-speed simulations can do billions of hands in a trice. Drawing conclusions from 20,000-hand samples is simply unacceptable and can and does lead to the kind of unreliable results you are experiencing.

Long-term, you will win counting and will lose playing BS (without bet variation). It's as simple as that. Of course, if you count to vary the bet, but play the hands according to BS, you will win, too, only not as much as if you used playing departures.

Win rates for playing alone as compared to with others vary very little for the shoe game. But, if you play alone, you will win much more per hour, because of the speed of the game. finally, the best approach to the shoe is back-counting. "Playing all" is both tedious and, often, unproductive.

My book, "Blackjack Attack," deals extensively with the shoe game and provides reams of statistics. So do several others.

Good luck to you.

Don

This is pure speculation but I recon that your standard deviation will be far higher playing head to head as you will play many more hands per shoe head to head than with say six players. The reason for this is that if a shoe has a large negative expectation (say with lots of aces and 10 hiding behind the marker) you will have to play the whole shoe yourself before it is shuffled again. With more people the shoe will be shuffled far more often.

As for being worse off head to head it may be that you are not worse off it is just that you have fallen on the wrong side of the higher head to head standard deviation in your tests so far.

Out of interest what player expectations do you get?

So in escence the results should be acurate.

I don't see why that follows. If you do another 20,000 hands, the results could easily (and probably would, actually) come out very differently. You might be led to the conclusion that counting 6d games, play all, is the road to quick retirement. Would THAT conlusion be, in essence, accurate? That's why we typically do hundreds of millions, or even billions, of rounds.

Think of your short run real world results as a snap shot, taken with a telescopic lense, of a tiny part of your climb up a very rugged mountain.

ETF

Most card counting advocates will tell you that most players (over 50%)will be in the plus column after 80,000 hands of play. If you hope to improve your chances of winning, MANY more hands must be played. If you want a 90% chance of winning as a card counter, you must play MILLIONS, maybe BILLIONS of hands....
Obviously, you can never be sure that you will profit as a card counter because you will never be able to play enough hands to reach long-term projections.
The same scenario exists for Basic Strategy and progressive bettors, yet card counters claim that BS and progressive bettors face a .5% disadvantage, while they have an advantage over the house.
Bottom line is that it's IMPOSSIBLE -- by the nature of the research -- to discount the merits of card counting.
I'd suggest that you play an additional 40,000 hands and see if you're any closer to making a profit.

"If you want a 90% chance of winning as a card counter, you must play MILLIONS, maybe BILLIONS of hands...."

You're just slightly off. The actual numbers, which vary slightly according to the game being played, are about 200 to 500 hours, implying 20,000 to 50,000 hands (see BJA, p. 24).

Bordeaux, you'll get a lot of good advice on this and other pages, but none of it will ever come from Walter! Be careful.

Don

Don said, "Drawing conclusions from 20,000 hand samples is simply unacceptable and can and does lead to unreliable results..."
I said, "If you want a 90% chance of winning as a card counter, you must play MILLIONS, maybe BILLIONS of hands."
He responded, "You're just slightly off. The actual numbers... are about 20,000 to 50,000 hands."
Sooo, in one case Don claims that 20,000 hands have no significance if you didn't experience a profit as a counter. In response to my post, Don claims that a counter has a 90% chance of winning with 20,000 to 50,000 hands...
Makes you wonder, doesn't it?

If I read Bordeaux post correctly he is stating that he is ahead slightly counting after 20,000 hands. Don states 20,000 hand is roughly 500 hours of play which could easily be done in a year's time.
So in a year's play, Bordeaux is reporting being ahead counting after 1 year of play. This falls in the 90% probability range. If he plays for 9 more years at the same rate, he will most likely see profit for 9 years with 1 losing year.

Of course the 20,000 hand sample is not large enough, the above post could easily be re-written with all the numbers reversed (the counting being slighly negative, the BS comparison being slightly worse, the # of hand winning streak/losing streak occuring exactly opposite, etc.) and it would not be stretching anything. It would just mean that that particular sample fell in the bottom 10% instead of the 90%.

He clearly states that he "found no advantage to counting in a six deck game over perfect basic strategy" and that "The results were within a few percent of each other in winnings with the slightly higher percentage in the non-counting sessions." In other words, Basic Strategy play was superior to card counting.
Don also stated that "Drawing conclusions from 20,000 hand samples is simply unacceptable.."
Bordeaux actually said that Basic Strategy play was superior to card counting for 20,000 hands of play, and that both systems were profitable. By your analysis, the player with the more positive record (in this case, the BS player) should win 9 out of the next 10 years. Is this correct?

That is not what I said. Please read Bordeaux's post at the top of the page. Please do not try to draw any conclusions from my posts. You are so clueless it is useless. Any attempt by me to explain anything to you is also futile. Please go back to the progression page instead of trying to twist and mis-read posts here into some kind of support for your nonsense.

Walter, you are so fundamentally deficient in your understanding of mathematics, statistics, and probability that I find it a disgrace that you write books and pass yourself off as an "expert."

Don

That if the 20K mark is stat significant for an advantage player (90% likely to be ahead), than it works similarly for a non advantage player (90% likely not to be ahead).

That would tend to depend on the betting style. The overall results of a card counter tend to be dictated primarily by a small number of double and split hands high in positive counts. A flat bettor or player using some kind of conservative progression will have a more consistent bet size. A martingaler's results, by contrast, are solely determined by whether that awful, inevitable negative streak stretching into the event horizon occurs sooner rather than later.

You would be measuring two different things, even if the overall number of hands for the different classes of players were the same.

Part of the problem here is that counters often refer to sample sizes of less than tens of millions of hands as "insignificant" or "meaningless". In fact, no sample size is meaningless. The smaller the sample size, the lower the level of confidence you would assign to any hypothesis supported by the data, but the data is never meaningless.

As an extreme example, Russell Barnhart in "Beating the wheel" discusses how to bet whilst clocking a roulette wheel for biased numbers. His recommendation is to bet on whichever number has come up most frequently - because that is likely to be the most biased number.
This holds true even for a single trial. In most instances this number will not, in fact, be biased, but this is the most effective method of limiting losses whilst collecting data, given that you believe that biased wheels exist at all.

I find it a disgrace that you write books and pass yourself off as an "expert."

most pros feel the same way about you.

I can always count on you for a remark that is so ignorant and baseless as to crack up the whole group.

Thanks for proving once again how totally CLUELESS you are -- as if we don't know whom we're dealing with here.

Don

"as if we don't know whom we're dealing with here."

I have my own suspicions as to who the poster is, but I want to know who you think it is. Your emphasis on the word "clueless" seems to be an allusion to GBV's "Baccarat for the Clueless." However, I doubt GBV posted that message.

Regards,
MF

No one else makes such moronic remarks about being a "real player."

Don

"But I am surprised that the 20K would be stat significant in any case. As you point out the confidence level varies. But at 20K I thought it would be so low that it would not be significant. I only perked up because Don indicated the 20k possible threshold."

You simulate a game playing millions upon millions of hands. You now know the EV and SD for that game. Now, I can make probabilistic statements concerning any number of hands, including 100. They're all perfectly valid. For example, I can state that, after playing 100 hands, there's, say, a 53% probability that the counter will be in the black. Then, to prove that, I play 100 hands, not once (of course!!), but millions of times, and -- lo and behold -- the counter comes out ahead 53% of those 100-hand trials.

In similar fashion, the page 24 BJA chart shows what happens for various game situations over various numbers of hands played, giving the probabilities of being ahead, based on the normal curve.

So, if you pick one of the better back-counted situations, you will see that, after only 20,000 hands played (200 hours), we have a 90% of being in the black.

I hope this is clear to everyone and that you understand the difference between what is being described here and what it means to run a sim for only 20,000 hands and to draw conclusions from that sim. I'm not talking about doing any such thing.

Don