How random is a RNG?

[daxter70]

throw the ole bell up there and lets see what ur sample and dist. looks like....

 

[Dorkus Malorkus]

There are many, many people who have analysed millions of hands between them and come up with nothing.

The Absolute threads demonstrate that if something weird was going on then it would get spotted by the number crunchers pretty quickly.

Just quoting this for emphasis.

 

[vanquish]

In theory, within a large enough sample size, there will be a stretch where someone, at some point, will be dealt AA 5 hands in a row, and they will be cracked each time. Basic probability ldo.

 

[Chris_TC]

In theory, within a large enough sample size, there will be a stretch where someone, at some point, will be dealt AA 5 hands in a row, and they will be cracked each time. Basic probability ldo.

That sample size would have to be, uh, very big. The probability of being dealt AA 5 hands in a row is 0.00000000019%

If your poker tracker contains roughly 526.3 billion hands then on average it should contain one instance of 5x AA in a row. Of course, getting them all cracked is a different story still

 

[DaFrench1]

There are many, many people who have analysed millions of hands between them and come up with nothing.

The Absolute threads demonstrate that if something weird was going on then it would get spotted by the number crunchers pretty quickly.


Is it just people doing this individually or can you point me to any websites where the numbers crunched and results can be perused?

Obviously, all the comments about sample size are correct, but at the same time if you only observe large samples you are also going to miss events at the micro-level. For example, If I run an analysis on 10,000 slot machine spins I might conclude that the machine has a 70% payout rate, but if I take the same sample and examine it in slices I might then see that at some stages the machine paid out 90% while at other times it only paid 10% but this would never have come to light when analysing the large number. I don't think its ever going to be easy to catch a site out because they can deal in very small margins (like we do as players). The other thing in their favour is what you all love to term as 'variance', I'm not denying its existence but what I'd like to point out is that as far as computers as concerened it is a programmable variable, and even shifts of small percentages away from the norm are going to make significant differences to site profits. And the biggest beauty of it is (if you are a site operator) is that you can (potentially) alter it when you want, or like the slot machine programme a floating variance (they pay out more when empty and less when full). At the moment it is all about trust, but you people seem to grant them that blindly yet are so analytical and questioning about other things. I find it a strange contradiction.

 

[ChuckTs]

Obviously, all the comments about sample size are correct, but at the same time if you only observe large samples you are also going to miss events at the micro-level. For example, If I run an analysis on 10,000 slot machine spins I might conclude that the machine has a 70% payout rate, but if I take the same sample and examine it in slices I might then see that at some stages the machine paid out 90% while at other times it only paid 10% but this would never have come to light when analysing the large number. I don't think its ever going to be easy to catch a site out because they can deal in very small margins (like we do as players).

This is what we call variance.

If you're flipping a coin, and it happens to land on heads 10 times in a row, is this because the coin is rigged? Of course not. It's because of statistical variance.

The really weird occurrences will happen an infinite number of times if you run infinite hands. If you run millions of hands, they will happen within the boundaries of statistical variance. The smaller the sample, the greater the variance will be. If you lose one hand with AA vs KK it doesn't mean a damn thing.

The other thing in their favour is what you all love to term as 'variance', I'm not denying its existence but what I'd like to point out is that as far as computers as concerened it is a programmable variable, and even shifts of small percentages away from the norm are going to make significant differences to site profits. And the biggest beauty of it is (if you are a site operator) is that you can (potentially) alter it when you want, or like the slot machine programme a floating variance (they pay out more when empty and less when full). At the moment it is all about trust, but you people seem to grant them that blindly yet are so analytical and questioning about other things. I find it a strange contradiction.

Slot machines have been proven to be programmed. Have poker sites? Nope. Sure, it's possible, but the fact that everyone's databases run within the boundaries of statistical variance means that we're nearly %100 sure they aren't programmed.

I mean no offense here, but I'm not sure you understand variance. Read wikipedia's page on it here.

 

[Dorkus Malorkus]

Why would someone rig a site in such a way that long-term expectancy is exactly the same as if the site were not rigged? What could a site possibly have to gain from this?

 

[ChuckTs]

heh, that too^^^

 

[Irexes]

Chuck's spot on with the coin analogy. In 100 tosses you will usually get a run of seven in a row, doesn't mean the coin's rigged like he says.

Taking 100 hands is just as bad.

At the moment it is all about trust, but you people seem to grant them that blindly yet are so analytical and questioning about other things. I find it a strange contradiction.

This is just patronising to a number of people here who have a very good understanding of stats and probability.

 

[iamthepush]

This was originally going to start out as a hand analysis thread but turned into something else once I retrieved my HH's from Pokerstars. Now, many people have been complaining about beats and saying that hands are set-up against each other and the like and they tend to get shot down pretty quickly. But tell me if this analysis of my last 100 hands on stars seems correctly distributed to you to be random.

Hand No. of times

AA - 2
KK - 2
Other PP - 8
AK - 4
A + Q,J,T - 2
K + Q,J - 4

Now that's 22 very decent hands in 100 (12 PP's?!). Now you're probably wondering how I wasn't the chip leader with so many nice catches, but you'll never guess what happened....

Now it could just be a freak occurrence, but I would request others to obtain their last 100 hands and analyse as above so we can get a better picture,

if you took 10,000 people and their avg weight was 185lbs and out of those people you took 100 at random and everyone indivual weight was above 200lbs, what would you do?

i would put them on a diet.

 

[Monoxide]

First, you have 14 RNG's and you determine the randomness of the randomest RNG then subtract 4 of the RNG's because they are rigged. Then add the sum of the randomest RNG to the least randomized adding into account the random chance your RNG changes its randomization.

 

[DaFrench1]

CHUCK T: Slot machines have been proven to be programmed. Have poker sites? Nope. Sure, it's possible, but the fact that everyone's databases run within the boundaries of statistical variance means that we're nearly %100 sure they aren't programmed.

I mean no offense here, but I'm not sure you understand variance. Read wikipedia's page on it here.[/quote]


Slot machines have not been 'proven' to be programmed, they were always 'designed' to be programmed as such. Hell, in the old days you would set the variance manually by flipping off the cover and turning a dial! Nowadays its all software controlled.

I have a science degree and understand variance. Do you understand illusion and deception?

Why would someone rig a site in such a way that long-term expectancy is exactly the same as if the site were not rigged? What could a site possibly have to gain from this?

My point is, as i have tried to illustrate, is that variance is a programmable variable as far as a computer is concerened (I also have a 12 year IT background so I know about this as well). Now you talk about number crunchers but I can't find any numbers or results that they are presenting, certainly not any cross-comparisons between different sites. You don't know if the variance you experience on sites is within norms or not, or even if the variance between different sites is the same or has wilder swings from site to site. There might even be times of the week or month when it is best to avoid playing because different algorithms maybe activated and running to achieve different outcomes. There are a loads of things that you don't know about how things are controlled and operated, but you all make out like you do...

Truth is there is no-one regulating this game and the potentials for abuse are only as limited as the imagination.

 

[Dorkus Malorkus]

You still haven't answered my question.

 

[ChuckTs]

I have a science degree and understand variance.

So you ask for a larger sample size to analyze, and when people throw tens of thousands of hands at you, you then talk about how the short-term results will yield more accurate proof? You obviously misunderstand statistics and variance then...

Please answer Dorkus' question:

Why would someone rig a site in such a way that long-term expectancy is exactly the same as if the site were not rigged? What could a site possibly have to gain from this?

 

[ratmantoo]

My 2 cents worth

After reading this post I appears to me that both "encampments" have valid arguments.

Yes - it is possible to adjust the "randomness" of events occurring. But why would a poker site want to do this as opposed to a slot machine. I know lots of slot players that will swear that a machine will pay out at a certain time / day / condition etc but not one poker player that says if i play in x tourney at y time Im going to win! (OK Im over exaggerating).

Yes - No computer is truely random. However I remember reading an article somewhere that the cards dealt are actually further randomised by events initiated by players on the entire site in addition to the deck being constantly shuffled - soz dont remember where but it was interesting.

Yes - Variance will occur . Take a coin and flip it 10 times - we expect a 50/50 heads/tails outcome but we seldom get that. Do this 10 times and then re analyse. each sample will not be the same but over the 100 flips we a likely to get close - over a thousand, a million even closer. percentage wise that is.

What DaFrench1 IMHO is trying to point out is by mini analysing the samples is that a tendancy for a certain hands MAY occur under certain conditions. In other words there may be a bias on the "randomness" of the hands at times


Well thats my 2cents and remember it South African cents so may not be worth as much as the American ones

 

[zachvac]

I think most accusations deal with that action cards are dealt out. Two things:

1. This would be difficult, they would have to program in something that adjusted the probability but then compensate in different times.

2. Although tougher, this is possible too. Check how many flopped flush draws you've had, and see how often you've hit. My offer still stands to run an sql query on my 20k+ hands and supply the query for anyone to run on their own.

 

[Monoxide]

After reading this post I appears to me that both "encampments" have valid arguments.

Yes - it is possible to adjust the "randomness" of events occurring. But why would a poker site want to do this as opposed to a slot machine. I know lots of slot players that will swear that a machine will pay out at a certain time / day / condition etc but not one poker player that says if i play in x tourney at y time Im going to win! (OK Im over exaggerating).

Yes - No computer is truely random. However I remember reading an article somewhere that the cards dealt are actually further randomised by events initiated by players on the entire site in addition to the deck being constantly shuffled - soz dont remember where but it was interesting.

Yes - Variance will occur . Take a coin and flip it 10 times - we expect a 50/50 heads/tails outcome but we seldom get that. Do this 10 times and then re analyse. each sample will not be the same but over the 100 flips we a likely to get close - over a thousand, a million even closer. percentage wise that is.

What DaFrench1 IMHO is trying to point out is by mini analysing the samples is that a tendancy for a certain hands MAY occur under certain conditions. In other words there may be a bias on the "randomness" of the hands at times


Well thats my 2cents and remember it South African cents so may not be worth as much as the American ones

0.02 ZAR

=

0.00304918 USD



lol

 

[ratmantoo]

0.02 ZAR

=

0.00304918 USD


Pathetic!!!! lmao

 

[blankoblanco]

Why would someone rig a site in such a way that long-term expectancy is exactly the same as if the site were not rigged? What could a site possibly have to gain from this?

don't worry, dafrench, i'll take this one. by clustering the bad variance together in chunks, the RNG induces more tilt from players than if it were more evenly spread out. more tilt equals more chasing losses/action equals more rake equals PROFIT. ldo

 

[Dorkus Malorkus]

Oh, I now consider myself a little more educated!

Also lol @ OP talking about running statistical analysis on small sample sizes to see if they're 'within expected variance' as if he's the first person ever to come up with the idea.

Guess how the AP scandal was brought to light? Oh yeah, by running statistical analysis on a few hundred hands where the results were so far from the expectancy even a small sample size of only a few hundred hands could be conclusive evidence of cheating. Guess what? No other similar findings on any other sites have ever been made (unless you're aware of something I'm not).

Really, stop throwing silly statistical nonsense that most of us are already aware of (you 'understand variance'? well whoopdeedoo, have a gold star!) around in an attempt to justify your baseless claims.

 

[DaFrench1]

So you ask for a larger sample size to analyze, and when people throw tens of thousands of hands at you, you then talk about how the short-term results will yield more accurate proof? You obviously misunderstand statistics and variance then...


I don't at all, what I'm saying is that you need to do at least BOTH, not to mention a load of other tests as well. Your argument to only analyse large samples is therefore just as invalid as someone who only analyses small samples .

 

[Dorkus Malorkus]

You still haven't answered my question.

*Tap tap*

 

[zachvac]

If you understand variance as well as you claim I'm sure you are also familiar with basic statistics. You can figure out the probability as this post did (https://www.cardschat.com/forum/general-poker-13/how-random-a-rng-96709/post-642607.html) and he found out there was a less than 2% chance. Now if this were a random set of 22 hands, you'd be right, but I doubt you would have posted this if you hadn't received the ridiculous run of cards. Here's your assignment:

Play 22 hands right now. Play 1/2 cent for all I care and just check/fold. Hell play the play money tables. Count your pocket pairs. The lower this probability is, the more of a chance it is rigged (although nothing can prove without a doubt that it's rigged, only that if it is perfectly fine, those results would be very rare, so we assume it's not if it's below a certain number). Do it and tell me how many PPs you get. That's how you use a small sample. The bigger the sample, the better though. Just make sure that before you test something you determine how many times you will do it and exactly what you are looking for, instead of making claims post facto. I could look at the fact that I got 37o, KJs, AA, 44, JTs, 39o, 28s and say the odds of that happening is (8/52)(3/51)(8/52)(1/51)(4/52)(3/51)(4/52)(3/51)(8/52)(1/51)(8/52)(3/51)(8/52)(1/51) = a very small number. And that's ignoring the order, if we take order and exact suit into account we can conclude that it's practically impossible. But that's because we did it post facto. If I make the claim beforehand that these cards will come up, and then they do, we can say there's an extremely low chance that this was due to chance, so we assume I am able to see these unfairly. Just some basic statistics that I'm sure you are familiar with if you understand variance as well as you do.

 

[jonfelkin]

I was playing live once and got AA 3 times within 7 Hands, different people delt the cards each time but i got the impression that something wasent right with it and they just wanted me to win a few hands as i was short stacked when i got it the second time (First time they got busted).

 

[MooseLee420]

random?

I do not think any random is entirely free of pattern language. Playing on-line, the potential for any sort of manipulation is HUGE, so we can only accept this as possible and probable, leaving us to trust in our game and the parts that will always be swayed by a pattern language.