Count correlation between shoes

Newbie counter here ...

Earlier I asked a few general questions, one of which was 'Is there any correlation between counts from one shoe to the next'.

The answer, generally, was 'NO'; the shuffle effectively removes any correlation so that, for example, in a 6-deck shoe where the true count ended at +2 the shuffle randonizes the deck such that there is no increased percentage of the next deck ending in a positive count.

Okay, this seems reasonable. However, in an article
( http://www.bjmath.com/bjmath/playing/tracking.htm ) I read that certain types of shuffles DO randomize the deck fairly well (i.e.: stutter shuffle) and others don't (i.e.: zone shuffle).

Is this article correct? If so, is the ramification for a zone-type shuffle that if the count ends very negative or positive, and depending on where the unplayed cards are placed back in the deck that SOME correlation can be assigned to how the deck will play out. For example, if the count ends very negative I know that the ratio of low cards in the unplayed portion is very high. If that the dealer uses a zone-type shuffle and that 'pile' (the unplayed cards from the previous show) is placed on top of the deck, and the cut card is placed in the middle or near the top of the deck, then can I assume that a greater ratio of high cards will be coming out the show near the beginning of the shoe. If so, I can use that information usefully. The opposite (where I know that the shoe will start w/ a high number of low cards) is that I can take a break, make a phone call, etc.

Or am I just on crack?

Thanks,
John

but you are talking about shuffle tracking, which is a whole new ball game from counting, in the sense that different skills are involved. You should read up on shuffle tracking. Can't help you there, as I am strictly a counter.

Neko

First you will have to determine what 'zone' that negative count unplayed portion of the cards you are interested in ends up in after the shuffle.

Then figure out what portion of the played cards it gets mixed with (and the count of that portion). Doesn't do any good to track a bunch of big cards, if they get shuffled together with a bunch of small cards, you end of with a section of 0 count.

Then you will have to place the cut card in the proper place to bring that section of the shoe into play, and know when you encounter that particular section when you play the next shoe.

Other than that it is pretty simple, isn't it?

If you have good reason to assume that the shoe will start with a lot of low cards, I would advise you to stay on highest alert because you will get very favorable situations later on in the shoe.
As a matter of fact that's what we are hoping for when we are counting the cards!

Francis Salmon

In most cases you won't be able to determine with what kind of cards a slug of high cards you have identified will be mixed together.
But you would be foolish not to use the slug information if you manage to determine where it ends up after the shuffle. If you know that a slug was I deck thick and was mixed together with about the same number of cards.If the slug originally consisted of 10 surplus high cards you are entitled to assume that this same slug which is now two decks thick will still have a surplus of 8 high cards (on average it will be shuffled together with a -2 portion).
This means that if you manage to cut this slug to the top, you will have a virtual TC of +4 over the first two decks. This might be worth trying.

Francis Salmon

You should be aware, if you are going to try this sort of thing, that most trackers lose, and I mean even extremely dedicated trackers who have trained very rigorously at this sort of thing.

Read R.W. Munchkin's interview with Johnny C. in the Fall Blackjack Forum, where he talks about the losses MIT experienced with various types of attempted tracking plays. The story has been the same on all the big teams.

It doesn't take much of a miss-cut to totally blow your edge. I have watched friends cut 3/4 of a deck in front of a slug and bet their hearts out into an 8% disadvantage, then drop their bets in discouragement just as their actual slug was about to start. It doesn't take much of this to wipe out a bankroll.

If the shuffle didn't affect flat shoes, flat counts on a shoe would stay that way for the rest of time. But they don't.

Doing anything with the type of shoe you describe would take the same types of skills as regular tracking, which, as Radar pointed out, are in short supply.

Evenness of distribution in what is seen implies nothing about evenness of unseen cards. The True Count tends to remain the same only as a gross average over many shoes. An individual shoe bounces all over the place. Otherwise, a shoe could never go positive or negative in the first place. And, once the shoe took a particular TC set, one could quit counting for that shoe and just bet and play. Microscopic inferences don't work. The correct answer for mid-shoe entry (without back counting) is as Wong has it in his book: count from entry and divide by the total number of decks unseen, whether before or after what you've seen. Makes for weak advantage play.

I had a different idea in the past that I should be able to walk up to a shoe in mid-deal, do a running count on a segment, and determine from a positive or negative value that the remainder of the pack was positive or negative (respectively) proportional to the count of that sample. I reasoned that, since a positive deck has a preponderance of high cards remaining, seeing a sample with a preponderance of high cards would indicate a preponderance of high cards in the remainder of the pack. It is true that if the deck is positive one will see a preponderance of high cards in any remaining sample (on average). However, the probability density function of a pack is so large compared to the results of the sample that one cannot infer that a sample with high cards means the rest of the pack has high cards. I had ignored a term in the Rule of Bayes.

I also might mention that I persisted in this notion until I wrote a simulation that proved myself wrong, primarily because I got mostly "hand-waving" arguments instead of mathematical explanations. Nobody said, "You've ignored a term in the Rule of Bayes." So answers you get may not be fulling satisfying.

All of the manual shuffles used for 6 and 8 deck games in Las Vegas (and Atlantic City) are imperfect, so there will be correlation in card order from shoe to shoe. Having said that, the degree of correlation is extremely slight.

Ignoring shuffle tracking concerns for the moment, the best way to characterize the "goodness" of the shuffle is to track some metric of non-randomness from shoe to shoe. I have chosen two metrics of interest: the probability of 10s following 10s (to please the clump trackers) and the difference between observed minimum and maximum running count. I find that for most shuffles employed in Las Vegas for shoe games, there is little correlation from shoe to shoe, but there is some. For instance, if the probability of 10s following 10s (P1010) is rather high for a certain shoe, lets say 40% (occurs less than 2% of the time), then the expected P1010 for the child shoe (after shuffling) will be around 32-33% for a stutter shuffle, which is noticiably higher than the "book" value of 30.55%. This is of academic interest only, as there seems to be no way to exploit these instances of deviations from non-randomness. A simpler shuffle is not as good as the stutter shuffle, but it still not exploitable in terms of changing overall strategy.

Shuffle tracking, however, is a differrent story, but this was not the subject of the post.

If you know that a slug was I deck thick and was mixed together with about the same number of cards.If the slug originally consisted of 10 surplus high cards you are entitled to assume that this same slug which is now two decks thick will still have a surplus of 8 high cards (on average it will be shuffled together with a -2 portion).

I don't understand this part. Where does the 8 come from? Could you please clarify?

Thanks,
Curious

I was always wary of over-reaching myself with tracking and tend to only use the information when it cannot or almost certainly will not harm my expectation. Cutoff tracking is a good example of this, it really can't hurt if you follow the count for playing and betting. I'm amazed other people don't minimize the risks in this manner, since most counters are by nature conservative.

It doesn't take much of a miss-cut to totally blow your edge. I have watched friends cut 3/4 of a deck in front of a slug and bet their hearts out into an 8% disadvantage, then drop their bets in discouragement just as their actual slug was about to start. It doesn't take much of this to wipe out a bankroll.

They should be factoring an error term into their advantage because of the imprecision associated with cutting. Pretending you can cut perfectly is suicidal and leads to instant ruin. The sort of cutting accident you describe will still happen if you use an error term but it will be in the context of an overall advantage, like betting into a high TC and watching the dealer get blackjack after blackjack, which balances out eventually.