vpFREE2 Forums

Interesting article

#1

There is an article on this subject by John Robinson in the current (March,
2004) Midwest Gaming & Travel Magazine (Mike Ditka on the cover) that makes
all of this serial vs parallel dealing mute. For about five years now
almost all VP machines continue to shuffle the remaining 47 cards (48 in
joker games etc.) while you make your draw decision. This is because under
the old methods somebody got hold of a VP machine and was able to figure
out exactly how the RNG genated it's numbers. They then used a computer to
determine where in the cycle the program was based on the five dealt cards.
They knew exactly what the draw cards would be and that works a heck of a
lot better than Winpoker or Frugal VP!!

···

[Original Message]
From: Yahtzee <junglekrma@yahoo.com>
To: <vpFREE@yahoogroups.com>
Date: 3/2/2004 7:23:24 AM
Subject: [vpFREE] Interesting article

For those of you who are not on Mark Pilarski's "Deal Me In" email
list, here's an interesting snip... enjoy.

-----------------------------
Dear Mark,
Let's say I am dealt four cards to a royal in video poker. I discard
one and
hope for the card that will give me a royal flush. My question is: is
the
card already behind the card I just discarded, or dealt from the top
of a
deck? Dylan F.
-
O.K., we've said it. Jurisdictional gaming regulations do vary, from
place
to place, but most poker machines operate on a "serial" and not
a "parallel"
basis. "Huh?" Dylan says aloud, while thinking "Whaaa"?
On the serial basis, cards are shuffled and dealt off the top of the
virtual
deck..just one after another as they come off the top. An example of
the
parallel basis, on the other hand, is where five cards are dealt face
up,
and five additional cards are then dealt unseen underneath the
initial,
face-up five.. parallel to them, get it?
So, Dylan, if it's serial, which it probably is, that next card is
determined at the time of player interaction with (draw from) the top
of a
virtual deck, and not at the time of the deal and card placement
behind your
five initial cards.
But either way you get those cards, Dylan, by gaming regulations in
all
legit casinos, they will be randomly drawn.
-----------------------

vpFREE Links:

http://www.west-point.org/users/usma1955/20228/VP/Links.htm

Yahoo! Groups Links

#2

Howard Stern wrote:

There is an article on this subject by John Robinson in the current (March,
2004) Midwest Gaming & Travel Magazine (Mike Ditka on the cover) that makes
all of this serial vs parallel dealing mute. For about five years now
almost all VP machines continue to shuffle the remaining 47 cards (48 in
joker games etc.) while you make your draw decision. This is because under
the old methods somebody got hold of a VP machine and was able to figure
out exactly how the RNG genated it's numbers. They then used a computer to
determine where in the cycle the program was based on the five dealt cards.
They knew exactly what the draw cards would be and that works a heck of a
lot better than Winpoker or Frugal VP!!

Heh, heh. But can you imagine the size of his strategy card. :wink:

Take, for instance, the parallel-method where there were 5 cards dealt underneath the 5 that are shown on the initial deal. With a 52-card deck there are 57,407,703,889,536,000 ways to deal those 10 cards. This is because in a truly random game, an JUNK hand like Ts 8h 7h 5d 3c must have 184,072,680 different combinations of the 5 cards that are underneath (sometimes there's a Royal underneath, and sometimes just another JUNK hand). So one would need to record the 5 cards dealt, discard them all, and record the 5 underneath, and that would presumably let you locate your place on the huge but finite sequence of hands that would be generated by an RNG.

But once the hand is located, it's gold from there on out. It's interesting to think about the best approach that could be used there besides sitting and sneaking looks at a computer generated list (which you would have to have) to decide your holds. The better approach would probably be to check the hands in all machines, go to your computer in your room and run them all, comparing them so you could pick the machine that is closest to providing the most opportunities for Royals, STFLs and QUADS -- and just have a short list of the important hands to play to get _those_ particular wins, and then just play the other hands using normal strategy. You could predict how long to wait before sitting down at that machine, and check it occasionally when it became available. That way you wouldn't be constantly looking at a list to attract attention, but you would be getting a very high percentage of big paying hands. Actually, on second thought, it might not be too good of an idea to be winning too many hand pays -- at least note by the same person on the same weekend. After you've hit your fourth Royal that weekend and the tapes show you looking at a piece of paper, I'd say that Guido is going to get a look at that paper one way or the other, and God help you if it's a list of holds for vp. :-/

So, was this a weakness that the casinos caught and fixed before it was ever exploited, or did someone actually do this? If so, what happened to them when they were caught?

Cheers.

Bill Velek

#3

Take, for instance, the parallel-method where there were 5 cards

dealt

underneath the 5 that are shown on the initial deal. With a 52-

card

deck there are 57,407,703,889,536,000 ways to deal those 10

cards. This

is because in a truly random game, an JUNK hand like Ts 8h 7h 5d

3c must

have 184,072,680 different combinations of the 5 cards that are
underneath (sometimes there's a Royal underneath, and sometimes

just

another JUNK hand). So one would need to record the 5 cards dealt,
discard them all, and record the 5 underneath, and that would

presumably

let you locate your place on the huge but finite sequence of hands

that

would be generated by an RNG.

But once the hand is located, it's gold from there on out. It's
interesting to think about the best approach that could be used

there

besides sitting and sneaking looks at a computer generated list

(which

you would have to have) to decide your holds.

Bill Velek

My understanding of VP machines was that the RNG was continuously
spining thru the possible shuffles of cards, and when the deal
button was pressed it stopped on the current shuffle. If this is
correct we would never know the following hand regardless of our
knowledge of the current hand. Can someone verify or correct this
info?

thanks
Jim

#4

My understanding of VP machines was that the RNG was continuously
spinning thru the possible shuffles of cards, and when the deal
button was pressed it stopped on the current shuffle.

Can someone please tell me why people continue to think this is important?

It keeps coming up over and over again.

If it's a random machine, it DOES NOT MATTER *when* the next card is picked,
so long as it's dealt out of a constant deck of cards.

For that matter, if it's a NON-RANDOM machine, it doesn't matter when the
card is picked, either, because you're just going to lose.

In either event, random or not, all your wishing, hoping, psychic
spoon-bending power or rosary bead twiddling isn't going to matter one iota
as to whether or not you're going to get the cards you need to complete your
hand.

#5

But once the hand is located, it's gold from there on out. It's
interesting to think about the best approach that could be used
there besides sitting and sneaking looks at a computer generated list
which you would have to have) to decide your holds.

Well, assuming that would be the case, and the casino wouldn't spot you
looking through either (a) a computer (illegal) or (b) a 3000 foot long
computer printout (bulky, but effective) and stop you, and assuming that in
that sequence of 57 X 10^17 hands you'd get no repeated deal sequences of
say 1000 hands that repeat in a pattern (entirely possible -- in fact,
probable) -- then you could up and down your betting until the good hands
came along. i.e. -- we know the next 4 hands are losers, so we bet 1 coin
each, and the next 2 are winners, so we bet max coin.

Yeah, like that's going to happen.

Your best hope would be that the series loops back in smaller, recognizable
chunks, and you could "Wong" a machine by preparing a few sequences to watch
for, then watching someone else play a machine, knowing they're getting to a
good point on the list, then wait until they get up and then jump in on the
machine.

Like a casino's going to let you stand behind someone and record their
play.... or that you could keep up with recording the play of a very fast
player, like me. :smiley:

Amusing speculation.

#6

jimnkelli wrote:

snip

My understanding of VP machines was that the RNG was continuously
spining thru the possible shuffles of cards, and when the deal
button was pressed it stopped on the current shuffle. If this is
correct we would never know the following hand regardless of our
knowledge of the current hand. Can someone verify or correct this
info?

I also thought that that was the case, and you are correct that such a constant shuffling would eliminate -- at least in my opinion -- the possibility of knowing what the next hand would be. I was merely responding to a post that suggested that it _used_ to be possible to know what was coming next because of the methods used at that time. If the desk was constantly shuffled back then, I would say that the original story would have to be bogus, because if you look at my example of a JUNK hand -- the Ts 8h 7h 5d 3c -- there would be no way to know to hold the Ts because there are Tens behind the 8h 7h and 3c to make Quads, unless the hands were being dealt in a know sequence based on the sequence of the RNG.

I'm not calling anyone a liar, but the whole thing is hard for me to believe because, if that really happened, the original computer programs were a bunch of idiots. Hell, it is so predictable that this would occur and it is so easily prevented (constant shuffling is just one of several easy ways) that Guido and the boys would be well justified in taking such incompetent programmers for a little ride.

Cheers.

Bill Velek

#7

jimnkelli wrote:

snip

> My understanding of VP machines was that the RNG was continuously
> spining thru the possible shuffles of cards, and when the deal
> button was pressed it stopped on the current shuffle. If this is
> correct we would never know the following hand regardless of our
> knowledge of the current hand. Can someone verify or correct this
> info?

I also thought that that was the case, and you are correct that

such a

constant shuffling would eliminate -- at least in my opinion -- the
possibility of knowing what the next hand would be. I was merely
responding to a post that suggested that it _used_ to be possible

to

know what was coming next because of the methods used at that

time. If

the desk was constantly shuffled back then, I would say that the
original story would have to be bogus, because if you look at my

example

of a JUNK hand -- the Ts 8h 7h 5d 3c -- there would be no way to

know to

hold the Ts because there are Tens behind the 8h 7h and 3c to make
Quads, unless the hands were being dealt in a know sequence based

on the

sequence of the RNG.

I'm not calling anyone a liar, but the whole thing is hard for me

to

believe because, if that really happened, the original computer

programs

were a bunch of idiots. Hell, it is so predictable that this would
occur and it is so easily prevented (constant shuffling is just one

of

several easy ways) that Guido and the boys would be well justified

in

taking such incompetent programmers for a little ride.

Cheers.

Bill Velek

I think some people were missing the point. Noone is saying that
they knew what was coming on the next hand or 700 hands from now.
All that they are saying is that they could decipher the last 5 cards
based on the result of the first 5 cards, for that hand only.

The way random nuber generators work are deterministic. That means
that if you know the algorithm and the seed numbers, you can
determine the next number. But what has always worked for video
poker makers is that the time between button pushes added enough
additional randomness that it was impossible to predict.

However, what it sounds like happened is that, in machines a decade
ago, the computer would grab the next 10 numbers off of the RNG as
soom as someone hit the deal button. Knowing the first five cards as
well as the algorithm could determine the next 5 cards down to a
small sample. But it won't help with the next hand, since some
random set of cards have gone by.

Now that they are reshuffling after the first 5 cards, the draw cards
are no longer the next off the deck, but some unknown millions of
cards later, preventing anyone from being able to predict the draw
cards.

···

--- In vpFREE@yahoogroups.com, Bill Velek <billvelek@a...> wrote:

#8

tooncesthecatwhocoulddriveacar wrote:

snip

I think some people were missing the point. Noone is saying that
they knew what was coming on the next hand or 700 hands from now. All that they are saying is that they could decipher the last 5 cards
based on the result of the first 5 cards, for that hand only.

I don't see how that would be possible because for every initial five cards -- whatever they may have been and regardless of their particular order -- it would be _absolutely_necessary_ that the would then be 47!/42!= 184,072,680 possible arrangements of the next five cards (for a 52-card) deck, ... otherwise the game would not be legal and would never have been approved by any state gaming commission in the first place. If we take the example I previously used -- Ts 8h 7h 5d 3c -- and that's what you saw come up as the initial deal, how could you possibly make any predictions based on just that unless the game was actually deliberately rigged with some sort of a backdoor combination by the writer of the program (something like a combination lock: hold all cards, then hold 0 cards ... and repeat those two holds 10 consecutive times and then a certain pattern will begin or a Royal will be dealt, or something like that). It would also be possible to program the machine to give a distinct advantage by playing a hand that no one else ever holds intentionally; for example, whenever you are dealt a Ts and a 3c, hold them both and discard all other cards and the next hand will be a Royal Flush. Either one of those examples are easily possible, but my comments were based on the description that was given, which I recall was that someone had obtained a machine and figured out its RNG sequence, or something to that effect.

In any event, none of this is going to happen to any of us, so I don't guess it really matters anyway.

Cheers.

Bill Velek

#9

Now that they are reshuffling after the first 5 cards, the draw

cards

are no longer the next off the deck, but some unknown millions of
cards later, preventing anyone from being able to predict the draw
cards.

So, If I understand correctly the current VP machines do
continuously shuffle untill the deal button is pressed, and then
they start shuffling the rest of the cards untill the draw button is
pressed. ??

that would mean that when you have been dealt 4 to a royal, when you
press the draw button is very important (not that you would have any
control over the outcome) as to whether you get the exact card you
would like to complete the royal.

I thought you always got the same cards off the top of the deck no
matter when you hit the draw button.

Jim

#10

All current IGT machines select the card(s) when the draw button is hit. It
is not generally known whether Bally, Sigma or other machines work the same
way, except that I have heard that there are jurisdictions that require
this method. If so, and if there are none that require 10 cards to be drawn
first, then certainly all machines made today would wait until the draw
button is pressed.

I thought you always got the same cards off the top of the deck no
matter when you hit the draw button.

Jim

[Non-text portions of this message have been removed]

#11

tooncesthecatwhocoulddriveacar wrote:

snip

> I think some people were missing the point. Noone is saying that
> they knew what was coming on the next hand or 700 hands from now.
> All that they are saying is that they could decipher the last 5

cards

> based on the result of the first 5 cards, for that hand only.

I don't see how that would be possible because for every initial

five

cards -- whatever they may have been and regardless of their

particular

order -- it would be _absolutely_necessary_ that the would then be
47!/42!= 184,072,680 possible arrangements of the next five cards

(for a

52-card) deck, ... otherwise the game would not be legal and would

never

have been approved by any state gaming commission in the first

place.

Actually, the requirements are that the cards be random (1/52 chance)
and unbiased (for example, a spade can't come up more ofen then
normal on a flush draw).

But let's say that the pseudo-RNG generates a stream of numbers is a
repeating sequence of 2^32 numbers mapped to the 52 cards. That's
4.3 billion numbers before a repeat happens. There are 312 million
permutations of the first 5 cards, since we will look at order as
well. That means that, on average, each hand will only be
represented 12 times in the sequence. And if there are only 12
possible sequences, there must be only 12 possible options for the
very next 5 cards. If you reverse engineered the code, you could
determine the 12 possibilities, and a computer (perhaps at home while
you talk on a cellphone) can determine the best option based on those
12 possibilities. I doubt that if the RNG were developed that way,
that it would have failed any randomness test that the gaming board
developed, since each card would appear with no detectable pattern
and each card would be represented by 82 million different random
numbers. That is, until the flaw was discovered, as it seemed to be
5 years ago.

···

--- In vpFREE@yahoogroups.com, Bill Velek <billvelek@a...> wrote:

#12

One more possiblity here. If pRNGs are "streaky", that is, they have
points in time where they would deliver a higher percentage of "good"
results. Then, someone could sit down at a machine with a phone to a
partner and they could determine whether the machine was in a good
cycle or bad cycle by playing a few hands. If in a good cycle,
continue to play. If in a bad cycle then move on. Over time this could
result in a significantly better pay back.

Dick

tooncesthecatwhocoulddriveacar wrote:

snip

> I think some people were missing the point. Noone is saying that
> they knew what was coming on the next hand or 700 hands from now.
> All that they are saying is that they could decipher the last 5 cards
> based on the result of the first 5 cards, for that hand only.

I don't see how that would be possible because for every initial five
cards -- whatever they may have been and regardless of their particular
order -- it would be _absolutely_necessary_ that the would then be
47!/42!= 184,072,680 possible arrangements of the next five cards

(for a

52-card) deck, ... otherwise the game would not be legal and would

never

have been approved by any state gaming commission in the first place.
If we take the example I previously used -- Ts 8h 7h 5d 3c -- and
that's what you saw come up as the initial deal, how could you possibly
make any predictions based on just that unless the game was actually
deliberately rigged with some sort of a backdoor combination by the
writer of the program (something like a combination lock: hold all
cards, then hold 0 cards ... and repeat those two holds 10 consecutive
times and then a certain pattern will begin or a Royal will be

dealt, or

···

--- In vpFREE@yahoogroups.com, Bill Velek <billvelek@a...> wrote:

something like that). It would also be possible to program the machine
to give a distinct advantage by playing a hand that no one else ever
holds intentionally; for example, whenever you are dealt a Ts and a 3c,
hold them both and discard all other cards and the next hand will be a
Royal Flush. Either one of those examples are easily possible, but my
comments were based on the description that was given, which I recall
was that someone had obtained a machine and figured out its RNG
sequence, or something to that effect.

In any event, none of this is going to happen to any of us, so I don't
guess it really matters anyway.

Cheers.

Bill Velek