vpFREE2 Forums

What Would It Take???

#1

OK, we'll do the mathematical proof thread on "Are VP machines
fair and random?" here on vpFREE rather than on FREEvpFREE.

Frank Kneeland's original post is appended below.

vpFREE Administrator

···

_______________________________________

From: Frank <frank@progressivevp.com>
Date: Sat, Mar 17, 2012 at 3:26 PM
Subject: [vpFREE] What Would It Take???
To: vpFREE@yahoogroups.com

A question that seems to be on a lot of people's minds is,
"Are VP machine fair and random?"

I answered this particular question to my own satisfaction
in the mid 90's when I had access to the enormous sample of
hands that being a team manager afforded me. I do not expect
anyone to believe me or take it on faith, and I'm a strong
supporter of skepticism and personal discovery, so...

It seems like every time someone, anyone, makes a comment
about machines not being random or merely shares their
beliefs and observations in forum, someone else (myself
included) chimes in and points out that their math is
flawed, their sample is too small, or their conclusion are
wrong. What almost never seems to get discussed is what
would be correct math, a large enough sample and a proper
method for reaching a good conclusion.

No one is going to be able to do it the way I did, so how
then???

I'd like to start this thread specifically for the purposes
of discussing what it would take to prove that machines are
not fair and random or visa versa as the case may be:

1. What should one record and how?

2. What is a large enough sample? (And how to parse the
sample.)

3. What math should they use to dismiss chance as the cause
of their results?

4. And perhaps most importantly how do they make the
information meet the standard of the scientific method, with
peer review and replication of results???

If #4 is not met any study done will only succeed in
convincing the person doing the study.

What's been offered so far hasn't convinced anyone that
wasn't already convinced and isn't likely to start
convincing people anytime soon. What we need is a simple
detailed template of the steps required that anyone can
follow and prove it to themselves, one way or the other.

I can think of no better way to spend our time on a VP forum
than working on this problem and resolving it once and for
all.

And please note: There should be no need for argument in
this thread about whether or not VP machines are fair and
random. That would be a conclusion. This thread is not about
conclusions, it's about how to formulate them.

---------------------

I've always found Bayesian Inference goes best with a nice
Chianti or Barbera...of course I'm always adjusting that
based on new information. ~FK

#2

Frank asked:

1. What should one record and how?
2. What is a large enough sample? (And how to parse the
sample.)

Record each type of card for each "play", for example I was dealt AdAc3h4h5h and drew 2hAhAs, that's 8 cards. Record 52 cards and you have a cycle. Record 260 cards and you have 5 cycles, which is statistically significant. At this point you are looking to verify (or disprove) that the average cycle of each card is 52. By 25 cycles you should have a fairly solid answer on whether or not any cards are missing from the deck. The next step is to look for correlation (stickyness or shuffle tracking) between cards. If the shuffle is truly random, the average cycle of each card should be 52 and there should be no correlation between cards, so for example the aces shouldn't be clumped together, or spread apart, anymore than would be predicted by probability theory.

http://wizardofodds.com/games/slots/jackpot-party/

#3

OK, we'll do the mathematical proof thread on "Are VP machines
fair and random?" here on vpFREE rather than on FREEvpFREE.

vpFREE Administrator

Here is the information from my GameKing for the two games we play the
most FPDW (100.76%) and KBJW (100.65%) includes both 5c and 25c
denominations.
In case someone wants to play with the numbers.

FPDW:
Amount Played $310,204
Amount Won $314,718
Machine Pay $305,968
Hand Pay $8,750

Game hold -1.45%
Game Yield 101.45%

Total games 248,578
Games won 112,052
Games lost 136,526

RF 6
4Deuces 63
RF/D 462
5K 816
SF 919
4K 16,115
FH 5,274
FL 4,231
ST 13,200
3K 70,966

···

----------------------------

KBJW:
Amount Played $594,358.75
Amount Won $604,349.25
Machine Pay $581,624.25
Hand Pay $22,725

Game hold -1.68%
Game Yield 101.68%

Games played 475,754
Games won 212,210
Games lost 263,544

RF 16
5K 51
RF/J 65
SF 250
4K 4,081
FH 7,397
FL 7,192
ST 7,008
3K 64,172
2P 52,727
K+ 69,251

Dennis
vp-connoisseur

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

#4

OK I love this idea. That would probably get us very sure very fast, but it would be laboriously tedious. You'd be playing at a snails pace and writing every single hand.

I have a question for everyone.

I think this would work, but how many would want to use it and would everyone prefer a less book keeping intensive method that tracked fewer things and allowed for more fluid play.

I'm not making this for me to use, so your opinions are more important than what I think.

~FK

···

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <nightoftheiguana2000@...> wrote:

Frank asked:
> 1. What should one record and how?
> 2. What is a large enough sample? (And how to parse the
> sample.)

Record each type of card for each "play", for example I was dealt AdAc3h4h5h and drew 2hAhAs, that's 8 cards. Record 52 cards and you have a cycle. Record 260 cards and you have 5 cycles, which is statistically significant. At this point you are looking to verify (or disprove) that the average cycle of each card is 52. By 25 cycles you should have a fairly solid answer on whether or not any cards are missing from the deck. The next step is to look for correlation (stickyness or shuffle tracking) between cards. If the shuffle is truly random, the average cycle of each card should be 52 and there should be no correlation between cards, so for example the aces shouldn't be clumped together, or spread apart, anymore than would be predicted by probability theory.

http://wizardofodds.com/games/slots/jackpot-party/

#5

That is indeed interesting info. One of my major concerns for this utility is that it be strategy independent. If one were to track made paying hands, strategy would influence the frequencies and bias the results.

I believe we have a way around this already, but thanks for the contribution.

~FK

--- In vpFREE@yahoogroups.com, vp-connoisseur <cacti-az@...> wrote: Here is the information from my GameKing for the two games we play the

···

most FPDW (100.76%) and KBJW (100.65%) includes both 5c and 25c
denominations.
In case someone wants to play with the numbers.

#6

For comparison here are the 2sd (95%) ranges at 475,754 hands for perfect play and random shuffle:
RF 11.5 +/- 6.8
5K 44.4 +/- 13.3
WRF 49.5 +/- 14.1
SF 273 +/- 33
4K 4066 +/- 128
FH 7457 +/- 173
FL 7410 +/- 172
ST 7895 +/- 178
3K 63434 +/- 504
2P 52862 +/- 460
KOB 67579 +/- 520

Prognosis: N0 for this game is 26.25/.006463/.006463= 628,436 hands so your dataset is below this figure, the end results are still more determined by luck than skill. Your straights are significantly deficient and KOB's overexpressed, probably indicative of play errors.

···

--- In vpFREE@yahoogroups.com, vp-connoisseur <cacti-az@...> wrote:

KBJW:
Games played 475,754
RF 16
5K 51
RF/J 65
SF 250
4K 4,081
FH 7,397
FL 7,192
ST 7,008
3K 64,172
2P 52,727
K+ 69,251

#7

OK I love this idea. That would probably get us very sure very fast, but it would be laboriously tedious. You'd be playing at a snails pace and writing every single hand.

Record it to video like the wizard did with jackpot-party. Analyze the results later. Also the video is your raw data, available for independent analysis by anyone.

···

--- In vpFREE@yahoogroups.com, "Frank" <frank@...> wrote:

I have a question for everyone.

I think this would work, but how many would want to use it and would everyone prefer a less book keeping intensive method that tracked fewer things and allowed for more fluid play.

I'm not making this for me to use, so your opinions are more important than what I think.

~FK

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <nightoftheiguana2000@> wrote:
>
> Frank asked:
> > 1. What should one record and how?
> > 2. What is a large enough sample? (And how to parse the
> > sample.)
>
> Record each type of card for each "play", for example I was dealt AdAc3h4h5h and drew 2hAhAs, that's 8 cards. Record 52 cards and you have a cycle. Record 260 cards and you have 5 cycles, which is statistically significant. At this point you are looking to verify (or disprove) that the average cycle of each card is 52. By 25 cycles you should have a fairly solid answer on whether or not any cards are missing from the deck. The next step is to look for correlation (stickyness or shuffle tracking) between cards. If the shuffle is truly random, the average cycle of each card should be 52 and there should be no correlation between cards, so for example the aces shouldn't be clumped together, or spread apart, anymore than would be predicted by probability theory.
>
> http://wizardofodds.com/games/slots/jackpot-party/
>

#8

OK that would work, wow great idea.

Are we sure that's legal to take video in a casino?

And two for people that don't have video cameras I'll still need to whip up a paper option, but thanks again.

Please someone comment on the legalities of this, I do not know the current laws.

Oh and we still need to answer the question of, "How many hands?". I'm shooting for 99% accuracy, if that's attainable. It may not be.

~FK

···

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <nightoftheiguana2000@...> wrote:

--- In vpFREE@yahoogroups.com, "Frank" <frank@> wrote:
> OK I love this idea. That would probably get us very sure very fast, but it would be laboriously tedious. You'd be playing at a snails pace and writing every single hand.

Record it to video like the wizard did with jackpot-party. Analyze the results later. Also the video is your raw data, available for independent analysis by anyone.

> I have a question for everyone.
>
> I think this would work, but how many would want to use it and would everyone prefer a less book keeping intensive method that tracked fewer things and allowed for more fluid play.
>
> I'm not making this for me to use, so your opinions are more important than what I think.
>
> ~FK
>
> --- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <nightoftheiguana2000@> wrote:
> >
> > Frank asked:
> > > 1. What should one record and how?
> > > 2. What is a large enough sample? (And how to parse the
> > > sample.)
> >
> > Record each type of card for each "play", for example I was dealt AdAc3h4h5h and drew 2hAhAs, that's 8 cards. Record 52 cards and you have a cycle. Record 260 cards and you have 5 cycles, which is statistically significant. At this point you are looking to verify (or disprove) that the average cycle of each card is 52. By 25 cycles you should have a fairly solid answer on whether or not any cards are missing from the deck. The next step is to look for correlation (stickyness or shuffle tracking) between cards. If the shuffle is truly random, the average cycle of each card should be 52 and there should be no correlation between cards, so for example the aces shouldn't be clumped together, or spread apart, anymore than would be predicted by probability theory.
> >
> > http://wizardofodds.com/games/slots/jackpot-party/
> >
>

#9

Alright, if the Admin is cool with it I'll hop back in. I want to make it very clear that I won't be part of denigrating anyone's beliefs, and that this thread should not be about people beliefs or conclusions.

It's strictly about the math one uses for testing a VP related hypothesis and methods for proof.

I would also like to be clear that I am pro skepticism, even if that skepticism is towards something I believe. I think that a measured amount of "do-it-your-self and don't trust what the other guy is telling you" is healthy.

When the project is finished, I'll post the method/utility for everyone to use themselves for themselves. What I'm not going to do is use the utility myself and then talk about my results.

If people share conclusions, then we open the door once again on doubt towards the accuracy of someone else's book keeping and agenda.

Do with it what you will, but I think it will be best as a personal tool, not a public one.

Sincerely, ~FK

···

--- In vpFREE@yahoogroups.com, vpFREE Administrator <vpfree3355@...> wrote:

OK, we'll do the mathematical proof thread on "Are VP machines
fair and random?" here on vpFREE rather than on FREEvpFREE.

#10

nightoftheiguana2000,

Awhile back I posted a file, Deal_Redeal.xls, in the "files" section of
vpfree.
This was 100 hands from a 10/7 DB game on my other machine.
I recorded a dealt hand, threw it all away and recorded the redeal.

Can this be used for any kind of analysis regarding the randomness?

Dennis
vp-connoisseur

nightoftheiguana2000 wrote:

···

--- In vpFREE@yahoogroups.com, vp-connoisseur <cacti-az@...> wrote:

KBJW:
Games played 475,754
RF 16
5K 51
RF/J 65
SF 250
4K 4,081
FH 7,397
FL 7,192
ST 7,008
3K 64,172
2P 52,727
K+ 69,251
   
For comparison here are the 2sd (95%) ranges at 475,754 hands for perfect play and random shuffle:
RF 11.5 +/- 6.8
5K 44.4 +/- 13.3
WRF 49.5 +/- 14.1
SF 273 +/- 33
4K 4066 +/- 128
FH 7457 +/- 173
FL 7410 +/- 172
ST 7895 +/- 178
3K 63434 +/- 504
2P 52862 +/- 460
KOB 67579 +/- 520

Prognosis: N0 for this game is 26.25/.006463/.006463= 628,436 hands so your dataset is below this figure, the end results are still more determined by luck than skill. Your straights are significantly deficient and KOB's overexpressed, probably indicative of play errors.

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

#11

Yea, if you happen to have a VP machine in your house that's the fastest way to do it. Of course I think people wouldn't take the results from a VP machine that wasn't in a casino, and I don't think re-drawing 5 cards every hand in a casino would be cost effective.

I guess you could find a penny machine and play only 1 coin. Every thousand hands would cost you not more than $10...and we'd get to hear all the fun stories about redrawn Royals, ye ha!

I don't think your idea is practical, but I'll keep it on the back-burner for now.

~FK

···

--- In vpFREE@yahoogroups.com, vp-connoisseur <cacti-az@...> wrote:

nightoftheiguana2000,

Awhile back I posted a file, Deal_Redeal.xls, in the "files" section of
vpfree.
This was 100 hands from a 10/7 DB game on my other machine.
I recorded a dealt hand, threw it all away and recorded the redeal.

Can this be used for any kind of analysis regarding the randomness?

Dennis
vp-connoisseur

#12

Most cell phones are now video cameras, I think the wizard specifically used a cell phone to record his video. Take a look at the wizard's page again, he's specifically teaching anyone how to do it:

http://wizardofodds.com/games/slots/jackpot-party/

Keep in mind, video poker is just another slot. The only difference is the specification of a random and fair shuffle of a 52 card deck.

As for 99%, why not go for 99.7%, that's 3SD. Approximately you'd need 10 cycles minimum, then 3SD is 10+/-9.5 . 10 cycles of 52 cards is 520 cards whereas 10 cycles of royals is over 400,000 hands, over 2 million cards. Obviously shuffle tracking the cards and not the end results is the way to go. The end results are just a derivative of the process.

···

--- In vpFREE@yahoogroups.com, "Frank" <frank@...> wrote:

OK that would work, wow great idea.

Are we sure that's legal to take video in a casino?

And two for people that don't have video cameras I'll still need to whip up a paper option, but thanks again.

Please someone comment on the legalities of this, I do not know the current laws.

Oh and we still need to answer the question of, "How many hands?". I'm shooting for 99% accuracy, if that's attainable. It may not be.

~FK

--- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <nightoftheiguana2000@> wrote:
>
> --- In vpFREE@yahoogroups.com, "Frank" <frank@> wrote:
> > OK I love this idea. That would probably get us very sure very fast, but it would be laboriously tedious. You'd be playing at a snails pace and writing every single hand.
>
> Record it to video like the wizard did with jackpot-party. Analyze the results later. Also the video is your raw data, available for independent analysis by anyone.
>
>
>
>
> > I have a question for everyone.
> >
> > I think this would work, but how many would want to use it and would everyone prefer a less book keeping intensive method that tracked fewer things and allowed for more fluid play.
> >
> > I'm not making this for me to use, so your opinions are more important than what I think.
> >
> > ~FK
> >
> > --- In vpFREE@yahoogroups.com, "nightoftheiguana2000" <nightoftheiguana2000@> wrote:
> > >
> > > Frank asked:
> > > > 1. What should one record and how?
> > > > 2. What is a large enough sample? (And how to parse the
> > > > sample.)
> > >
> > > Record each type of card for each "play", for example I was dealt AdAc3h4h5h and drew 2hAhAs, that's 8 cards. Record 52 cards and you have a cycle. Record 260 cards and you have 5 cycles, which is statistically significant. At this point you are looking to verify (or disprove) that the average cycle of each card is 52. By 25 cycles you should have a fairly solid answer on whether or not any cards are missing from the deck. The next step is to look for correlation (stickyness or shuffle tracking) between cards. If the shuffle is truly random, the average cycle of each card should be 52 and there should be no correlation between cards, so for example the aces shouldn't be clumped together, or spread apart, anymore than would be predicted by probability theory.
> > >
> > > http://wizardofodds.com/games/slots/jackpot-party/
> > >
> >
>

#13

Out for rest of day.

I have no idea if what I'm attempting is even possible or what the end result will be, and like most good science, "research is what I'm doing when I have no idea what I'm doing."

Thank you all for your ideas, keep em-coming and I'll post a "what's left to figure out" in the morning.

~FK

#14

The first entry jumps right out at me:

TcQd3c2dKc Kd2cKc3sQs

You redrew the Kc?

Possibly an entry error, another reason why video is so important.

A script could be used to check for redraw errors. Then the totals could be gathered, there's 1000 entries, so the 95% expectation is that each card occurs 19.2+/-8.8 times. You could also calculate the chances that a particular card would not occur in 100 draws of 10 cards, that would lead to the odds of a card missing from the data by chance alone. That's a good question for the wizard: in 100 draws of 10 cards from a randomly shuffled deck, shuffled before each draw, what are the odds of a card missing from the results by chance alone?

···

--- In vpFREE@yahoogroups.com, vp-connoisseur <cacti-az@...> wrote:

nightoftheiguana2000,

Awhile back I posted a file, Deal_Redeal.xls, in the "files" section of
vpfree.
This was 100 hands from a 10/7 DB game on my other machine.
I recorded a dealt hand, threw it all away and recorded the redeal.

Can this be used for any kind of analysis regarding the randomness?

Dennis
vp-connoisseur

nightoftheiguana2000 wrote:

>--- In vpFREE@yahoogroups.com, vp-connoisseur <cacti-az@> wrote:
>
>
>>KBJW:
>>Games played 475,754
>>RF 16
>>5K 51
>>RF/J 65
>>SF 250
>>4K 4,081
>>FH 7,397
>>FL 7,192
>>ST 7,008
>>3K 64,172
>>2P 52,727
>>K+ 69,251
>>
>>
>
>
>For comparison here are the 2sd (95%) ranges at 475,754 hands for perfect play and random shuffle:
>RF 11.5 +/- 6.8
>5K 44.4 +/- 13.3
>WRF 49.5 +/- 14.1
>SF 273 +/- 33
>4K 4066 +/- 128
>FH 7457 +/- 173
>FL 7410 +/- 172
>ST 7895 +/- 178
>3K 63434 +/- 504
>2P 52862 +/- 460
>KOB 67579 +/- 520
>
>Prognosis: N0 for this game is 26.25/.006463/.006463= 628,436 hands so your dataset is below this figure, the end results are still more determined by luck than skill. Your straights are significantly deficient and KOB's overexpressed, probably indicative of play errors.
>
>
>

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

#15

I never noticed that before. I would have to think it was my typo error. Then that means the data is corrupt!

Thanks,
Dennis
vp-connoisseur

nightoftheiguana2000 wrote:

···

The first entry jumps right out at me:

TcQd3c2dKc Kd2cKc3sQs

You redrew the Kc?

Possibly an entry error, another reason why video is so important.

A script could be used to check for redraw errors. Then the totals could be gathered, there's 1000 entries, so the 95% expectation is that each card occurs 19.2+/-8.8 times. You could also calculate the chances that a particular card would not occur in 100 draws of 10 cards, that would lead to the odds of a card missing from the data by chance alone. That's a good question for the wizard: in 100 draws of 10 cards from a randomly shuffled deck, shuffled before each draw, what are the odds of a card missing from the results by chance alone?

#16

Cool plan if 5 cycles is significant. Appealing to me because my feel is that the NSU games in OK sure seems to throw out a lot of 2 Pair hands. In Vegas it seems more rare and kind of a good hand when you get it. In OK the sink the boat. But that's just my feel. I'm ready to start taking notes. But if somebody out there already knows that they games are not random here, they could save me some money by telling me before I start this.

···

________________________________
From: nightoftheiguana2000 <nightoftheiguana2000@yahoo.com>
To: vpFREE@yahoogroups.com
Sent: Monday, March 19, 2012 12:13 PM
Subject: [vpFREE] Re: What Would It Take???

Record each type of card for each "play", for example I was dealt AdAc3h4h5h and drew 2hAhAs, that's 8 cards. Record 52 cards and you have a cycle. Record 260 cards and you have 5 cycles, which is statistically significant.

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

#17

This kind of STUFF is why I will be leaving this group. Seems some people want to always run the show and hog the web space.

This whole subject is drivel and serves little purpose in my opinion!

···

--- In vpFREE@yahoogroups.com, "Frank" <frank@...> wrote:

Out for rest of day.

I have no idea if what I'm attempting is even possible or what the end result will be, and like most good science, "research is what I'm doing when I have no idea what I'm doing."

Thank you all for your ideas, keep em-coming and I'll post a "what's left to figure out" in the morning.

~FK

#18

Frank wrote:
... One of my major concerns for this utility is that it be strategy independent. If one were to track made paying hands, strategy would influence the frequencies and bias the results...

···

===================
Frank,
I think that a study like this (involving results) is the best:
http://wizardofodds.com/online-gambling/blr/

Keeping track of things like completing full houses from dealt two pairs, completing quads from dealt trips, and completing royals from dealt four to a royal can help us decide whether machines are random or not (these are the things that lead us to believe they aren't).
Just keeping track of how many times a card appears over so many cycles seems like a fairly easy programming way to gaffe a machine. For example, if I need 7c to complete a quad and don't get it, but it pops up in the next hand, and this happens 150 times in a row...I'm suspicious about the randomness, but the card has appeared right on schedule. The Wizard did a test like this on the 5Dimes website a few years ago, but I can't find it now.

Speaking of making a royal from dealt 4RF, does anyone have info on the Western Wisconsin casino (about 20 years ago) that got caught changing chips? The programming was done so that if you actually were going to complete the royal, another card would be drawn, so that the probability of completing the a royal was 1/47^2.

#19

Major positive development.

I just found a friend that is a visual basic programming expert with skills far beyond anything I've ever seen, that will work for food. OK gourmet comps, but hay it's still food. This will greatly improve the usability and functionality of the final product.

Once we lock down what we want coded, he can do it in the blink of an eye. I'm meeting him tonight.

~FK

#20

As far as I knew literally every single video poker player in the world has a view on one side of the random vs. non-random argument. And currently a lot of people are undecided or unsure. I thought a simple to use utility that helps people quantify their results and be more sure would be welcomed by all. I must say I'm at a loss to understand your objections or why you think the subject is drivel.

I hope you are wrong, because if you are right, I'm wasting quite a lot of my time for free.

If it makes any difference to your not leaving the group, I'm getting close to having enough info to begin the project, and in perhaps less than three days will closing the thread to begin work on the utility.

Check back in next week and things should be back to normal.

~FK

--- In vpFREE@yahoogroups.com, "gamblinman46825" <aesop31@...> wrote: This kind of STUFF is why I will be leaving this group. Seems some people want to always run the show and hog the web space.

···

This whole subject is drivel and serves little purpose in my opinion!