Hi Guys, does anyone out there know anything about shuffle machines, specifically the 'Shuffle star' and the "king' shuffler, is there a way to beat them?
I know of some casino's that are low on security that could be taken for a lot if it is poss to beat these machines.
Forget it, they cannot be beaten.
Dear Cynic, I actually know of a team that is currently doing well, I need help so that we can do the same and have a share. (remember, in this game anything is possible, don't be too quick to dismiss something just because it is unfamiliar to you).
The two types of machines are 'Shuffle star' and 'Quick draw' not 'king' as in the origonal posting (I really should post things in the morning and not after a marathon session) ;o)
Are these machines the continous shuffle machines, or simply an automatic shuffler. If the cards are placed in it after a shoe is done and removed to play the next shoe (typically w/ new decks of cards) the game is beatable. But if the cards are continually placed into the machine and played out of it, it is like playing against an infinite deck of cards...and therefore not beatable by counting. I'm no expert by far, this is just information i've picked up from bj21.
bac
If there indeed was a way to beat these machines, discussing it on an open message board that is certainly read by casino personnel would not be very smart now, would it?
I've watched the guts of the standard ShuffleMaster machine, which produces a *very* good shuffle. Since it's more "random," it should be easier to beat than the *far* less than adequate manual shuffles the machines replaced, at least in the multi-deck games. MUCH less like-card bunching -- only as much as would be normal in a more or less "random" situation.
The ShuffleMaster King is a CSM, which produces a very strange effect -- an effect I cannot prove mathematically, but have observed over and over again.
Each multi-deck shoe goes through what I call "phases" -- a dumping phase, a choppy phase, and a "the-damned-dealer-never-seems-to-break" phase. These phases can be long or short, and can occur in any order, as would be expected in a relatively "random" shuffle.
But the King machine, in addition to being uncountable, is also fatal to the normal basic strategy player, as the machine somehow manages to greatly extend the length of the various phases. When you get into a "choppy" phase, it seems to last FOREVER. This is survivable if you know how to bet into a choppy shoe, but for most players it means a painful death. The "dealer-never-breaks" phase also seems to last FOR-F**KING-EVER, and this is not survivable no matter how you bet.
But the "dumping" phase is also greatly expanded, and if you can catch one of these tables during one of its exceedingly rare dumping phases, you can make a LOT of $$!
The problem is that the table will sit empty for hours on end because the $%*&@% machine is in a choppy or cold phase; so there is little or no play to advance out of the bad phase to get you into a good phase. I am convinced this is the reason so many of these machines have been taken out of the casinos -- they ain't no revenue when the table sits idle.
If there is a way to beat them, the LAST thing you want to do is even mention it in public. You should consider deleting the thread, and asking your questions privately.
A CSM does have, with usual settings a skewed latency where when a card does not appear it likely has been placed deep into the pack and is more likely to remain undealt than an actual continuous shuffle would provide. A CSM that did such a shuffle would have a distributed latency that would provide a very narrow edge IMHO. Banger effects are amplified by the way it inserts discards randomly. Those bends are more likely to cause like cards to cling together.
But the biases you describe are the same as those claimed by the TARGET system and are wrong!
Clumping, as examined by Wong in the 1995 editon of Profesional Blackjack actually HELP the player. Even if you ditch elements of TARGET like the cigarette butt count, TARGET is almost exactly wrong.
You are correct that there are ways to beat CSMs but they are nothing to get excited about, and you are WRONG about TARGET type concepts being the ways to do it.
I am not familiar with the "target" system. I've heard rumors that it involves stuff like table selection criteria, etc., but know little else about it.
I've also heard the theories that CSMs should provide a slightly higher edge to the basic strategy player, and it sounds great, in theory -- *until* you observe (or worse yet, play at) an actual table with a ShuffleMaster King machine on it.
As far as like card bunching goes, some years ago -- back in the mid to late '80s -- I wrote a model of the stutter shuffle game using the then current AC rules and procedures.
With 5 players at the table, the computer-random-shuffle sim showed a slight advantage in 6-deck game using the hi-lo count. The model, which dealt, picked up, stacked, and shuffled the cards *exactly* the way the AC dealers did, showed a virtually even game (a house edge so slight as to be statistically questionable).
Using basic strategy only, the sim came up with about a 0.55% house edge, but the house edge in the model was about a full 1%.
Still skeptical about like-card-bunching theory, I tried to think of ways to examine the shoes after the respective shuffles to detect it. I decided the ultimate and most easily detectable like-card bunch would be a pair -- e.g., two 3s or two 9s, etc., in succession.
Starting at 10 cards below the "cut card" and going to the top of the shoe, each newly shuffled shoe was checked for adjacent identical value cards. The stutter shuffle model came up with approximately 3% more pairs than the CRS sim (which did behave the way the math would lead you to expect).
I have my theories about how and why this occurs, but the bottom line is that an inadequate shuffle (such as the stutter shuffle, where the top and bottom � decks are only riffled twice and the middle 5 decks riffled 3 times -- before the cut, of course) does produce more like-card bunching.
I also read where someone did a study of a *single* deck where the deck was not shuffled at all, and this helped the player. To this, I say: so what? This exercise has nothing to do with what actually goes on at a pretty much full table using a 6-deck stutter shuffle, and therefore proves exactly nothing. It would produce *dramatically* different degrees and types of like-card bunching.
Fortunately, stutter shuffle games have been pretty much replaced by automatic shuffling machines.
I have modeled these as well, and the results show no difference than a sim using a computer random shuffle. It is on these results that I based my statement that the ordinary ShuffleMaster machine produces a *very* good shuffle.
I have not yet modeled the ShuffleMaster King machine, because I do not know how they work internally -- i.e., how the locations are determined for the insertion of the used cards. I *could* make some assumptions based on how I would design such a machine, but such assumptions could be worse than useless. So, absent any specific knowledge of the guts of the SM King, it is not possible to create a reliable model.
If you have information about how the SM King determines where to insert the used cards, please pass it along. I will then create a model of it and post the results here. I will not only be looking at overall results, but on the lengths of both winning and losing periods aggregated for 5 players. IOW, if the aggregate for 5 players is rising, this would be considered a "dumping" phase, and if falling this would be considered a "killing" phase. Since all would be flat betting playing basic strategy, a period with a more or less flat aggregate would be indicative of a "choppy" phase.
about 2 months ago. Set the wayback machine, the search limits, back as far as possible and check.
This is the one time and probably one time only that you can find a situation where a progression might work. It is only because the machines only do random inserts of the discards into the pack to be dealt. This means that the average delay in having a card re-appear is proportional to the square of the depth of the insertion. Card put 90% back in the pack have (.9/.1)^2 or 81 times as long to wait to re-appear as cards put 10% back in the pack. A random number generator decides where to put the discards.
Mr. Harris quickly found an anti-cut-card effect with his modeling of CSMs but no progression effect. I theorized this would come from where the cards sent deep had low cards, and would result in slight gains from a progression, but only because this squared skew distribution overturns normal assumptions about latency. That is because a delayed card is likely to remain delayed, which opens the possibilities to streaks being more probable. But such bias, leading to a player edge, are not only going to take longer to disappear, due to this squared skew delay, but they are also going to take longer to form. I suggested that posters then try composing decks to see if results, "tend to hang," and to date no one else has tried it, and I am reluctant to post my results.
But the TARGET recomendations are still almost exactly wrong, even for this strange situation. There is no predictable series of phases; the only bias is that the normal Bayesian assumptions about depleated populations simply reverse themselves. There is no pattern to when and to what degree a pack with a bunch of low cards will be sent deep. All that can be said is that getting rounds with more high cards is more likely to follow getting rounds with more high cards, than usual. What does show up, and fairly early at that, is that a discard round made up with low cards is larger, and when sent to the same depth as any other round will tend to result in pushing forward the other cards more, relatively, and this anti-cut-card effect builds slowly.
Other than these two trends there is no pattern. The rest of what you allege is pure TARGET. You yourself revealed why your sims and other studies tended to support this: you only examined how like cards clinging to like cards resulted in more pairs. That is not what happens with banger damaged cards. The low cards can have two diagonals, while the tens and aces endup with opposite diagonal bends, that are opposite to the low cards, in whether the bends are up or down ones. The types of cards clump exactly as was the case in Wong's no-shuffle sims. The clumping is not absolute and your conclusions are false as a result. These are tendencies and not as exact and determinant, as you indicate you think, as well.
I even had some data on this I failed to preserve from a former sky employee at the Atlantis Resort in Reno Nevada. He had taken tracking software files from there, and discovered other casino's files on some hardware that had been backed up from other casino and actualy found a .2% rise in basic strategy players gains (at least they were tagged as basic strategy players or better by his filtering scripts) when peak devices were combined with normal shuffle machines. That is perhaps the biggest data error I ever made.
>But such bias, leading to a player edge, are not only going to take longer to disappear, due to this squared skew delay, but they are also going to take longer to form.<
This conforms *exactly* to my observations.
Obviously, the "take longer to form" part clearly implies the crappy phases of the shoe will last longer as well.
>There is no predictable series of phases;...<
Wait a minute. I never said they are "predictable!" I merely said they occur. Jeez, if they were predictable, I'd own the entire Gulf Coast *_*_*_*_*
With the info you have supplied, I will now construct my "model" and see what I get in the way of overall results and the duration of "phases." Thank you *very* much.
I did a search (2 months max allowed) and couldn't find the posts you spoke of.
However, the info re random insertion is all I need for now.
Thanks again.
You certainly speak like a true progressionist though. I more or less wanted to show how counters could "jump on these landmines for us," in agregate, so that casinos would find CSMs too unprofitable to keep around. But then just try to find a progressionist that plays accurate basic stratregy. I have one tactic that can train them fast.
Check this link
I got to all these posts by setting back 4 months. Try again please!
Thanks.
In reading the thread, I've come up with one more question:
Is the insertion process similar to riffling the discards into the remaining cards, but with the number of remaining cards between inserts determined by the RNG?;
OR
Is each discard to be inserted placed randomly *anywhere* within the remaining cards?
I think I know the answer, but am not absolutely sure of it; so further clarification would be appreciated.
Thanks.
One of the machines just uses a series of trays and does not riffle the discards in. I forget whether this is the Quick 5 or the King. There is no spread between card that is set by the RNG.
In Minnesota here they have Shuffle master machines. They are not continous shuffle machines. I love to beat them. The reason a continuous shuffle machine is unbeatable is not because of card sequences, card orders, randomness or nonrandomness it is because the cards are put back. The game then is no different than dice or any other casino game. The reason we are able to beat them without replacement is because favorable conditions occur while certain denominations of cards are removed without replacement. When those conditions occur we milk the table. Of coarse if you put the cards back every time you never have favorable conditions.
