Randomness of Shufflemaster

Are Shufflemaster machines totally random or are they pseudo controlled by the casino similar to slot machines limiting the number of paying hands to whatever the casino wants? I've noticed several things that make me ask this question... the fact that the machine keeps track of every card (it can tell the dealer what card is missing), and that often, the card that would complete a big hand is in the adjacent hand. Also, the machine 'knows' how many players are at the table, as the dealer presses his 'complete' button after all the hands have been dealt. This post deals with Three Card Poker and Let it Ride.

Thanks. If this is a duplicate post, I apologize... please point me to the original.

Quote: phd

Are Shufflemaster machines totally random or are they pseudo controlled by the casino similar to slot machines limiting the number of paying hands to whatever the casino wants? I've noticed several things that make me ask this question... the fact that the machine keeps track of every card (it can tell the dealer what card is missing), and that often, the card that would complete a big hand is in the adjacent hand. Also, the machine 'knows' how many players are at the table, as the dealer presses his 'complete' button after all the hands have been dealt. This post deals with Three Card Poker and Let it Ride.

Thanks. If this is a duplicate post, I apologize... please point me to the original.

Hi! Start reading here and to the end (6-8 posts) for a very detailed discussion of what SHFL machines do in various modes.

Quote: phd

Are Shufflemaster machines totally random or are they pseudo controlled by the casino similar to slot machines limiting the number of paying hands to whatever the casino wants?

One more post on the subject:

Shufflemaster

Of course they are rigged. That's why the casinos pay such exorbitant lease fees.

Dealing with a single deck and 3 card hands, I think there should be about a 1 in 8 chance that a particular card that isn't in your hand will be in the hand to your left or right. Higher if we're talking about 2 outs instead of 1 (like a high or low complement of a straight flush, or the third of a kind). (Someone with stronger math could grind it out.)

I have yet to see any "totally random" shuffle, ever.

"totally random" shuffles are a very vaugue concept. It depends on what the machine is meant to do.. Match house shuffle, random number of cuts and mixes, perfect shuffle large piles, etc. Machines also have to be careful, as too many perfect shuffles could leave clumps of cards in the same order as they were when they were put in, leaving one who is good at tracking cards to get the advantage.

tricky concept IMO

Quote: ajemeister

Machines also have to be careful, as too many perfect shuffles could leave clumps of cards in the same order as they were

That's riffles, not shuffles.

As far as I know, elevator type machines (like an MD2 or MD3) do a Fisher-Yates shuffle. They're as random as the random number source that drives them. Any card in the starting deck can end up in any position in the shuffled deck.

My understanding is the wheel type machines (like a one2six or iDeal) do a variation on a Pile shuffle. My understanding is the method used is almost as good as a Fisher-Yates shuffle. It should also be much faster.

Machines like the Shuffle Tech ST1000 do a series of operations to mimic riffles and strips. This should be about as good as a typical (non-professional) hand shuffle. (A typical professional dealer hand shuffling will often incorporate an offset break or two, which should keep cards from getting "stuck" at the bottom or top of the deck.)

Inexpensive machines like this do a single riffle per pass, and in my experience, do it quite badly. In order to be effective, several passes would be required, as well as one or more strips and offset breaks. (These are also slow, noisy, and prone to jamming and flipping cards.)

I haven't gotten a good look at a DeckMate (or DeckMate2) to figure out what it does. I'm guessing it's an elevator shuffler.


While I haven't ever seen a totally random shuffle, some of the shuffles I've seen on poker tables are very very good. On most house banked games, I generally see weaker shuffles employed as a concession to time.

Quote: Dieter

As far as I know, elevator type machines (like an MD2 or MD3) do a Fisher-Yates shuffle. They're as random as the random number source that drives them. Any card in the starting deck can end up in any position in the shuffled deck.

Dieter, I understand Fisher-Yates as a shuffling algorithm. Do you know what the MD2 or MD3 uses as the RNG?

For context, I've been creating my own non-interactive simulator for fun to generate indexes on a SP21 variant and I'm getting different numbers than expected. I've been through the code a couple times, but I'm not seeing any logic errors so I'm trying to get the process as close to real life as possible. I'm wondering if it even matters since, if you run enough iterations then in theory, you'd run into the same iterations, just at different times.

Quote: style1001

Quote: Dieter

As far as I know, elevator type machines (like an MD2 or MD3) do a Fisher-Yates shuffle. They're as random as the random number source that drives them. Any card in the starting deck can end up in any position in the shuffled deck.

Dieter, I understand Fisher-Yates as a shuffling algorithm. Do you know what the MD2 or MD3 uses as the RNG?

For context, I've been creating my own non-interactive simulator for fun to generate indexes on a SP21 variant and I'm getting different numbers than expected. I've been through the code a couple times, but I'm not seeing any logic errors so I'm trying to get the process as close to real life as possible. I'm wondering if it even matters since, if you run enough iterations then in theory, you'd run into the same iterations, just at different times.
link to original post

I realize this thread is 10 years old but my question is for Dieter and what makes him think they are using a Fisher-Yates shuffle?

Quote: DRich

Quote: style1001

Quote: Dieter

As far as I know, elevator type machines (like an MD2 or MD3) do a Fisher-Yates shuffle. They're as random as the random number source that drives them. Any card in the starting deck can end up in any position in the shuffled deck.

Dieter, I understand Fisher-Yates as a shuffling algorithm. Do you know what the MD2 or MD3 uses as the RNG?

For context, I've been creating my own non-interactive simulator for fun to generate indexes on a SP21 variant and I'm getting different numbers than expected. I've been through the code a couple times, but I'm not seeing any logic errors so I'm trying to get the process as close to real life as possible. I'm wondering if it even matters since, if you run enough iterations then in theory, you'd run into the same iterations, just at different times.
link to original post

I realize this thread is 10 years old but my question is for Dieter and what makes him think they are using a Fisher-Yates shuffle?
link to original post

Once upon a time, I was very curious about what happened inside a shuffler.
From everything I heard, it was a Fisher-Yates.

I probably read a few patents, and all the marketing material I could get to.

It was 10 years ago. A lot has happened since then. If you know better, I'm interested to learn.

Quote: style1001

Quote: Dieter

As far as I know, elevator type machines (like an MD2 or MD3) do a Fisher-Yates shuffle. They're as random as the random number source that drives them. Any card in the starting deck can end up in any position in the shuffled deck.

Dieter, I understand Fisher-Yates as a shuffling algorithm. Do you know what the MD2 or MD3 uses as the RNG?

For context, I've been creating my own non-interactive simulator for fun to generate indexes on a SP21 variant and I'm getting different numbers than expected. I've been through the code a couple times, but I'm not seeing any logic errors so I'm trying to get the process as close to real life as possible. I'm wondering if it even matters since, if you run enough iterations then in theory, you'd run into the same iterations, just at different times.
link to original post

I don't know what they're using as an RNG.
I expect it's good, or at least good enough to pass a GLI review.

Random is random.
If you're simulating a black box that randomizes an input deck to an output deck, I don't think it matters which method you shuffle with - if both shuffles are good.
If one of the shuffles isn't good, and you want similar results, only then do you need to mimic the bad shuffle.


Now, with all that out of the way, I'm assuming that the RNG embedded inside is an implementation of whatever was generally regarded as state of the art 5 years before the devices hit the market.
I wouldn't be surprised if MD3 is new enough to have Mersenne Twister, and the MD2 is a generation before that... but this is speculation.

Quote: Dieter

It was 10 years ago. A lot has happened since then. If you know better, I'm interested to learn.

No, I do not know any different. I just thought it seemed strange that any company would disclose the method they are using.

Quote: DRich

Quote: Dieter

It was 10 years ago. A lot has happened since then. If you know better, I'm interested to learn.

No, I do not know any different. I just thought it seemed strange that any company would disclose the method they are using.
link to original post

This makes sense.
Can I ask, do you know a lot of working programmers that would rather come up with their own method instead of adapting (mostly copying) a working solution out of Knuth?
(I don't, but I don't hang out with a lot of programmers these days.)

So the assumption that they're not reinventing the wheel is part of my reasoning.

Quote: Dieter

Quote: DRich

Quote: Dieter

It was 10 years ago. A lot has happened since then. If you know better, I'm interested to learn.

No, I do not know any different. I just thought it seemed strange that any company would disclose the method they are using.
link to original post

This makes sense.
Can I ask, do you know a lot of working programmers that would rather come up with their own method instead of adapting (mostly copying) a working solution out of Knuth?
(I don't, but I don't hang out with a lot of programmers these days.)

So the assumption that they're not reinventing the wheel is part of my reasoning.
link to original post

I don't recall working on a gaming RNG that was textbook. They were all slightly modified or a combination of multiple algorithms. Also, once a company had one they liked they rarely changed it. Some I worked on were 20 years old.

to me the most basic and random shuffle that i know goes as such...

you have two piles of cards... a pile can contain 0 cards at first.

you take a random card (position) from the first pile
and then put that card into the second pile.
restart. until pile 1 is empty

this from what i understand can recreate all possible shuffles available in the universe.

Quote: DRich

Quote: Dieter

Quote: DRich

Quote: Dieter

It was 10 years ago. A lot has happened since then. If you know better, I'm interested to learn.

No, I do not know any different. I just thought it seemed strange that any company would disclose the method they are using.
link to original post

This makes sense.
Can I ask, do you know a lot of working programmers that would rather come up with their own method instead of adapting (mostly copying) a working solution out of Knuth?
(I don't, but I don't hang out with a lot of programmers these days.)

So the assumption that they're not reinventing the wheel is part of my reasoning.
link to original post

I don't recall working on a gaming RNG that was textbook. They were all slightly modified or a combination of multiple algorithms. Also, once a company had one they liked they rarely changed it. Some I worked on were 20 years old.
link to original post

can you name individual shuffles that may have been in combination with each other? if not i understand... maybe they dont even have names come to think of it

Quote: heatmap

Quote: DRich

Quote: Dieter

Quote: DRich

Quote: Dieter

It was 10 years ago. A lot has happened since then. If you know better, I'm interested to learn.

No, I do not know any different. I just thought it seemed strange that any company would disclose the method they are using.
link to original post

This makes sense.
Can I ask, do you know a lot of working programmers that would rather come up with their own method instead of adapting (mostly copying) a working solution out of Knuth?
(I don't, but I don't hang out with a lot of programmers these days.)

So the assumption that they're not reinventing the wheel is part of my reasoning.
link to original post

I don't recall working on a gaming RNG that was textbook. They were all slightly modified or a combination of multiple algorithms. Also, once a company had one they liked they rarely changed it. Some I worked on were 20 years old.
link to original post

can you name individual shuffles that may have been in combination with each other? if not i understand... maybe they dont even have names come to think of it
link to original post

Sorry, I can not remember any specifics. I do remember some used one RNG to seed the next RNG. Honestly, I don't know that there is any merit in that.