Shuffle Machines - 103
(More continuing Education about Card-shuffling Devices used in Blackjack)
by Ron Fitch --- a.k.a. Boris
Since 1995, I have been writing articles on the subject of Shuffle Machines. Originally, there were only 2 ShuffleMaster units and an odd assortment of machines made by other manufacturers, that are now (sadly) long out of operation, as players found ways to exploit design flaws in the shuffling procedures utilized by those machines.
As I have said in the previous articles, thanks to information provided to me by knowledgeable ShuffleMaster spokespeople and the patent diagrams found at http://uspto.gov, I have been able to provide you with an inside look at how these shuffle machines actually work and how (in most cases) we as players can take advantage of their design "features". Rather than retire most of these machines, casinos have simply tried to switch them into different game venues, giving us yet another opportunity to exploit their weaknesses.
Beginning in 2000, ShuffleMaster has begun to introduce a series of what are known as "Continuous Shufflers". The idea behind the continuous shufflers is that the idea of "penetration" (so important to card counters) has all but been elimenated. According to the manufacturer, these shuffle machines are designed to produce truly "random cards", rather than actually non-random cards. In the past, units like the ShuffleMaster MD-1 (the "BOX" Shuffler) have made claims to the production of random cards, but in fact often produce quite non-random cards, exploitable to some card counters, and clump-trackers in particular.
As I write this article, I have just implemented ShuffleMaster's KING & One-2-SIXtm shufflers in the Boris for Blackjack software. In the last two weeks I have had the opportunity to play Blackjack against these simulations, finding them to be considerably different from the shuffle machines that have come before. Nevertheless, with a few modifications to Basic Strategy I have been able to show a noticeable profit in playing against software simulations. When some of my other duties are safely behind me I will spend time with Boris' Strategy Design facility to test these variations against multi-million round simulations.
The KING Shuffler
The KING Shuffler (and it's single-deck counterpart, known as the "ACE), utilize a card-sorting method which involves the use of "Shelves". Because these articles are being written from the Blackjack perspective, I will contain my focus to only the KING shuffler. I will then introduce the One2Six shuffler and its carousel method.
Shelf stacking is accomplished by stripping off single cards from an input stack and inserting then one at a time on of a number of "shelves". Then, as needed, each shelf is randomly unloaded into the dealing-shoe mechanism. Currently, only ShuffleMaster uses this method. According to the patent diagrams, the KING shuffler sports 19 card-stacking shelves.
What is interesting about the number 19 is that it is a prime number. The use of prime numbers in random number generators reduces the chance of "duplicate" numbers being generated from different formulas (sometimes known as "collisions"), as opposed to say, 18 (divisible by 2, 3, 6 and 9) or 20 (divisible by 2, 4, 5 and 10). As a side note, being a professional software developer, I get a kick out of ShuffleMaster's marketing "smoke" about their use of [sic] 64-bit random number generation, when in fact, they are generating only two entries per card-insertion: a value from 1 to 19 for determining which shelf a card should be placed on, followed by another value (essentially 1 or 2) to determine whether the card is placed on-top .vs. underneath any cards already on the shelf - you need 64 Bit number generation for that?
Because each shelf can hold a maximum of 14 cards, the KING shuffler best supports either 4 or 5 decks; at the discretion of the casino manager. I have a theory that with 5 decks, the possibility of cards "jamming" on shelves is more likely than with 4 decks. Do the math:
5 Decks (260 cards) / 19 Shelves gives us 13.6842 cards per shelf (not counting cards in the dealing shoe)
4 Decks (208 cards) / 19 Shelves gives us 10.9473 cards per shelf (not counting cards in the dealing shoe)
So as you can see, with 5 decks, most of the shelves will be nearly full most of the time. It is a fact that early KING shufflers used to jam a lot; which might explain why dealers were instructed to accumulate more than one round of cards in the discard tray before stuffing them back into the machine. The more rounds in the discard tray, the more likely we might be able to exploit what is known as latency of redistribution.
In my talks with players and casino management alike it is clear that early KING shufflers had some serious mechanical problems. In the 2000 - 2001 era, it was not uncommon to see a unit with its covers off; which brings me to an important point.
The only write up on shuffle machines in a generally available Blackjack was written by Jerry Patterson. Overall, Jerry's Blackjack: a Winner's Handbook (2002 Edition) is an excellent re-release of his 1990 Edition, sporting a chapter on Shuffle Machines, which graciously gives me and Boris a nice plug, for which I am grateful. Unfortunately, this chapter is riddled with errors, and is arguably the most questionable chapter, in what is otherwise a largely well-written book. This chapter on shuffle machines gives a completely erroneous description of how the KING shuffler operates.
Jerry writes up alleged observations by someone known as "007". Being a long-time Boris software user, Mr. "Bond" and I have traded numerous e-mail exchanges over the years, leaving me with nothing but respect for his Blackjack acumen. Unfortunately, regarding the KING shuffler, he has it all wrong; which makes me wonder what he was smoking while on the deck of that cruise ship.
Bond talks about the cards being "clamped off" and eventually moving down in sequence toward the bottom. Who knows, maybe that was an early (now discontinued) prototype of the machine, but that is not how the KING operates, if you study the patent diagrams and descriptions of the unit.
Aside from those mistakes, Patterson's comments on exploiting a machine near the end of its "cleaning cycle" may have some merit. It will take some careful record-keeping (as part of your table scouting) in order to discover this, because unfortunately, simulating a [sic] "dirty" shuffle machine is beyond my ability as a software developer. It is one thing to program in the effects of humidity on a hand shuffle; it is yet another domain entirely to program in the effects of wear and misalignment in a mechanical shuffle mechanism.
Now, depending upon your point of view, the introduction of the ShuffleMaster KING has had a devastating-effect on the game of Blackjack. Because of the way in which the cards are inserted on the shelves from an input hopper, KING essentially introduces the concept of continuous-shuffling but without any risk of latency of redistribution.
shuffling is LOADED with pros and cons.
With the KING shuffler, the dealer can put discards into the input hopper at any stage of the game, creating in-effect a continuous-shoe. Because there is no shuffle-card, card-counting success is virtually eliminated. Card-counters require "penetration" in order to jump their bets. I guess we could say that card counters are in effect looking for a sort of latency of redistribution, which is not possible with properly executed shelf stacking.
With the KING shuffler, as long as the dealer inserts the discards back into the machine every round there is no possibility of an exploitable latency. In a 7-player game, if two-rounds or more are left in the discard tray before being reshuffled, the latency of redistribution problem may again resurface.
Jerry Patterson quoted me as saying that I thought the machine may be producing clumped cards. While I do remember suggesting that early on in my observations, having since observed a multitude of KING games (and now the machine simulations under Boris for Blackjack), unless there is something I am missing, cards produced by this unit are essentially random - clumping simply does not exist.
As for the cons to this shuffler, I have already mentioned the mechanical problems, although they may well be a thing of the past. As it turns out, most dealers hate the continuous shufflers. Hand shuffling gives the dealer a break from the tedium of the game, lost with a continuous shuffler; but what do casino managers care - dealers are just minimum-wage (plus tips - paid by the players) employees.
While the KING shufflers have gained in popularity (casinos like the PALA Indian casino in California use SMBOX for their green/black chip games and a bevy of KING shufflers at their lower stakes Blackjack tables) other casinos (such as the Venetian in Las Vegas) have removed the KING units altogether, while leaving the BOX shufflers in play; probably because high-roller players don't trust continuous shufflers.
It will take a carefully run simulation with Boris for Blackjack to tell us whether there is an exploitable latency with these units. For more information on the subject of latency of re-distribution idea, consult Chapter 9 (Buck Rogers Blackjack - machine vs. counter) of John May's book Get the Edge at Blackjack (Pp. 79-87).
While I contest the application of May's material against the KING shuffler, ironically, like with Patterson, his material may still have application elsewhere. However that is for another article.
Part of the success of ShuffleMaster Corp. has been in buying out the rights to various games and card shuffling products. Some years ago, the bought they rights to the QuickDraw shuffle machine, although ironically, it seems as if they bought it to pull it OFF the market, while keeping the rights to a number of its patents. In more recent times they have struck a collaborative effort with CARD-Casinos Austria Research and Development to combine some of their patents with a "carousel shuffler device to produce what is known as the ONE-2-SIX shuffler.
While physically, the KING and ONE-2-SIX units look quite different (inside and out), in fact they embody an almost identical card-sorting idea. While the KING utilizes 19 shelves, the ONE-2-SIX sports 40 slots (altho the patent diagrams show 50). In essence, the ONE-2-SIX is simply a bunch of vertical shelves curved to form a circle - like a ferris wheel. Each slot holds only 10 cards (instead of 14), but otherwise the principle is largely the same. ShuffleMaster claims the unit can house 1 to 6 decks, although theoretically it could hold up to 8 decks.
For people looking to exploit latency of redistribution, the ONE-2-SIX is probably not exploitable in that regard at all. Like the KING shuffler, this is also a continuous shuffler, rendering card-counting and clump-tracking virtually worthless in these games. As to whether or not high-rollers will play against these machines, time will tell.
We have seen shuffle machines evolve of the last 15 years, for better or for worse. While they no doubt increase a casino's hands-per-hour, if players refuse to play at tables with these machines, considering their high monthly rental/maintenence costs, they could turn out to be a bad investment.
Something that casino managers need to consider is this: while hand-shuffled games can be exploited by knowledgeable players, against average players (the majority) they tend to produce a considerably larger house edge for casinos who employ them (depending upon which studies you believe this can be as much as a 2% - 5% increase). A larger edge means greater profits for the casino's bottom-line. Additionally, hand-shuffling has been shown to reduce dealer "burnout"; although as they are minimum-wage employees, quite possibly casino managers could care less.
If the continuous shufflers do indeed produce TRULY random cards, then the casino edge over all players will be reduced back to under 1 percent and most of the Blackjack books which promulgate Basic Strategy will again become relevant. It may again become a wise investment for players to invest $3 or so in a Basic Strategy "crib-card"; unless casino's again disallow their use at the Blackjack tables. We could argue that the casinos will make up the difference in "volume". However, I seriously question whether they can derive enough "volume" to offset the difference.
Like it or not, continuous shufflers are here to stay. However with software simulations like we now have in the Boris for Blackjack computer software, astute players may well find these continuous shuffle machines to be the best thing that ever happened to the game of Blackjack.
Latency of Redistribution
According to Get the Edge at BLACKJACK (P. 80) John May describes "latency of redistribution" as it applies to continuous shuffle machines as follows:
In layman's terms, this means cards played on one round probably won't appear on the next. Exactly when the cards find their way back into the shuffle depends on the number of players, the speed of play, and most importantly the variety of machine itself.