Gnaw on this... its about the rigged thingy

Ok I agree that a site deals umpteen thousand of hands. However it calculates hands per the "whole site", it doesn't calculate hands per individual tables. Think I may be wrong, I may be right wrong.. more then likely I am wrong.. Just stop and think for a moment..

A pair is supposed to happen 16:1 and specific pair is 220:1

So theoretically those are supposed to happen once in 16 hands you play or 220 hands you play.

However the rng isn't going to know this.. why is that?

Well the more tables it plays the more hands it will deal..

Now it deals with a whole site of hands compared to individual tables.. so instead of dealing with 1 table with individual hands it deals with multiple table with multiple hands.

Just like if there is per say 20 video poker machines running, of course now your chances increase of a royal flush happening compared to 1 machine running.

So if an rng could seperate individual tables compared to a whole site then could you get correct data on why unbelievable hands come more often then they happen.

Another point is to be made , you are playing a multi tourney and your table is diminished after say 10 hands. You are going to have to start the %'s of hitting someting over when you are transfered to a new table.

The original table you were at no longer exsists and so don't those hands but since an rng won't know this, it will continue to generate hands based apon hands that are no longer in existance.

Now I don't know how many people are dealers out here but I am sure they don't see that many spectacular hands very often just like a certain table shouldn't see a spectacular hand very often.

I do agree that we do see more hands quickly online then in real life but I do also believe that we should seperate the "individuality of a table" just like casino's do.. but not make it a whole casino like a site does... cause I am sure as a "whole casino" it sees more spectacular hands compared to the individual tables who hardly see many spectacular hands.

But then again I still and will always believe there are much much much more loose aggresive players online then in real life.


Just a "foolish theory"... toss it in the air.. and giving out guns to all those who wish to shoot it down.. lol even myself

I was wondering some of the same things this week.

Is there a program out there that will take your hand histories and tell you if it's been actually random ?

Naw.

I agree that a site's randomness must be for the whole site and NOT per table. I believe it because when I play 2 tables at once on any site, I get the same hand, different suits quite often. If my two hands aren't the same, the flop on one table is usually similar to my hand on the other. It gets to where I can almost predict what's coming. Doesn't help because it's never AA. LOL.

OK I'm trying to understand your theory, but am getting nowhere; I'm just going to post my thouhgts after a few quotes:

Wha? The way I see it, a real RNG doesn't calculate 'per' anthing...it just chooses random cards for every person's holding, of course making adjustments for other people's holdings (ie not dealing the same card twice).

Exactly Chuck, the RNG, is actually a PRNG, which stands for Psuedo Random Number Generator. It's merely an algorithm that spits out numbers. Each number cooresponds to a card in the deck. Once that number is used in a hand it is no longer available for uses again until the start of a new hand. The reason it is pseudo and not actual is that the algorithm is a formula and given the same variable to start with, it will always give the same result. If 4+C=a then whenever C=5, a will always = 9. In this case, the letter a is the initial variable, or seed. Of course the algorithm for generating poker hands is longer, but not much more complex. The randomness comes from the seed. In our example above, the letter C performed that function, and we chose the number 5 at a whim. Computers don't have whims so there the seed must be chosen carefully, changed continuously, and carefully guarded. This is done in several ways, some sites use a timer and the seed is merely a number chosen by say the time of day, or the time since the hand started, or whatever. This is considered weak, since the time of day isn't a national secret. Others use natural more unpredictable stimuli such as the position of your curser on the screen. SO, anyway, the PRNG could give a rats ass whether you've gotten your share of AA or 72o or what table your at. Over 13000 hands I've been dealt 93s only 21 times and A4s 54 times. Both should be the same. Why aren't they? Because it's RANDOM!
Incidently, I've hit AA once out of every 256 hands. I've been robbed.
Now look what you've done! I'm late.

Terry Pratchett [author] once wrote:

A million to one shot happens 9 times out of 10

How do you know these facts? I manually went through one of my MTT HH's to see the distribution and prove to myself that I really WAS getting bad hands. It took forever to do and that was only 115 hands or so.

PokerTracker.

Gnaw has got a G in it.

ok...still trying to understand your theory, twizzy (man i was tired last night....coulda sworn it was MrS posting it )
a certain percent (6.7%) of a player's hands will be pairs. As the number of people at a site increases, more pairs will be dealt - but not a higher percent (theoretically). For a site with 10,000 hands dealt, the number of pairs dealt will be ~667. For a site with 1,000,000,000 hands dealt, the number of pairs dealt will be ~66,666,667. Sure, alot more pairs were dealt in the second example, but in both examples the percent of the total hands that were pairs is the same.

So from what i've got from your theory: You think that the RNG has mixed up the numbers and the # of pairs dealt at your single table with the # of pairs dealt at the whole site? And thats why you get so many pairs?

....ok I'm still confused :/

You can test whether you're getting too few or to many aces or any other hand using the following:

Take the number of sample hands you have (N) and multiply it by the probability of getting that hand (eg 1/221 for pocket Aces). Call the result L. This is the mean of the distribution (expected number).

The variance of the distribution is also equal to L.

To get your test statistic z, take your actual number of that hand and subtract L. Then devide this by the standard deviation, which is the square root of the variance, i.e. sqr(L).

Now decide on a significance level, e.g. 1%, and look the significance limits up on a Normal (z) distribution table such as this one. To save you the bother:

At 1% the limits are -2.58 < z < 2.58
At 5% the limits are -1.96 < z < 1.96

So, from my database, under one of my aliases I got AA 23 times in 7487 hands.

L = (1/221)*7487 = 33.88 (I should get on average this many)

z = (23 - 33.88)/sqr(33.88) = -1.87

At 5% significance, -1.87 is within the limits above, so I reject the hypothesis that the rate I'm getting aces is not equal to (1/221). In other words, although I'd expect to get Aces 34 times but only got them 23 times, this doesn't show that I'm not actually getting them at the "proper" rate.

Try it yourself for your stats! Don't forget that even if they are complying to the correct mean value, 1 time in 20 on average they will be outside the 5% significance bands (that's why they're 5% significance).

Particular pocket pair - expect 1/221
AK suited - 1/332
AK offsuit - 1/110

More probs here

As an aside this is as a Normal approximation to the Poisson distribution. In reality we should use a Binomial distribution since the hands are discreet, but the approximation is close enough. For slightly better accuracy use variance = n*p*(1-p) rather than just n*p, but as p is small, it doesn't make a lot of difference. (Trying on the example above gave z = -1.85 rather than -1.87.)

I see the same thing alot, but there are only 52 cards. When you see two tables have similar flops, they usually aren't the same suits. I have seen it quite a bit, but its just the way the cards fall. There are only 13 different cards (forget suits for now) to fall. You are bound to see twop flops at the same time that may 2 or even all 3 of the same values. I really don't think the original poster is on to anything. If some online poker player can think of this I am sure that the programmers have thought of this as well.

Exactly.

Exactly my point it could care less what you've been dealt.. it only deals with the entire site and not indivudual tables.. so a descrpency can be determined that it only deals with the entire site and not indiviudual tables..

So the computer can't carefully have whims so the program is only meant to run for an entire site and not indivdual tables.. hence table #1 that is on a tourney table that only plays 4 hands and is discontinued so to even out the tourney tables. It will only place C from what it knows and that is all the hands being dealt from all the tables and not individuale tables.

The outcomes will change if you play the same table compared to multiple tables.

If you play table X 1000 times and you play table a,b,c,x 1000 times the outcome won't be the same

As espected an rng only valuates the outcomes of # of hands compared to the outcome of expected hands per whole tables.

Now if I play 4 tables my espectations are increased to hit a specific pair compared to 1 table... Just like an rng it only calculates that same expecation per the whole site compared to individual tables.

I don't think the poker sites calculate expectations at all, for individual tables or the entire site. They just apply a shuffling routine based on random numbers to the deck before each hand. The random numbers are created by some kind of clever maths which takes various inputs including "system entropy", i.e. readings CPU and memory usage, clock time and date etc, but nothing to do with the way other deals have come out, on that table or across the site.

So the way the cards come out for each hand is random. Your expectations of numbers of types of hands are just based on the laws of probability - nothing to do with the software. You'd get the same distributions if you dealt physical cards by hand, so long as you shuffled well.

Twizzy, I'm not sure what you're getting at, but I think you're assuming that the PRNG is somehow trying to spread out certain outcomes so that everyone gets the same even distribution of expected cards. But even though, that might SEEM fair, it wouldn't be random at all. A truly random distribution always includes a number of statistical anomalies. I would be highly suspicious of any site where my distribution was too perfect over a statistically meaningless period. Now, the more hands I play, the more thing should even out, but some events are so rare, a Royal Flush for example 649,750:1, that you could play your entire life and not expect to see one. Or you might get 10.

If you notice though, they wash the cards way more then they shuffle. Shuffling actually bends the cards believe it or not.. I personally can't be delicate enough with a deck, I for some reason always shuffle them aggresively. Yes they don't bend or the corners bend when I first shuffle but with time they will.. some just take longer then others. Hence why you will see more washing the cards then shuffling.

It has to spread out for the "entire site" it doesn't do it for "individual tables".. would the same suspicions be included where it was not perfect say 10% of your hands were good hands over a meaningless period.

Should and could are 2 diffrent things.. agreed about the royal flush. Now there is only 52 cards in a deck period and multiply that by who knows how many tables are playing at 1 site. The rng has to take X from 1 hand and put it somewhere else randomly, so X can also be in someone elses hand on Table Q. Yet Table Q can see that same card cause it wasn't randomly done from the same table which is dealt lesser hands.. Table Q can get Table E's card that is the same value as X.

Random has to be done from 52 cards only on 1 individual table not X number of hands from the whole site done from X number of tables.

It only generates a number based apon how many hands, not how many individual tables have dealt the number of hands.

If table Y is .01/.02 cent tables it will generate many many many more hands compared to table W that is 100./.200 dollar tables. Now when put together table Y and table W you get more generated hands compared to individual hands per table.

Um? Yeah....I guess.

You've really lost me on this thread Twizzy. What makes you think each deck of cards isn't shuffled on a per table basis? You'd have to know the details of how the RNG works to say it doesn't. Which site in particular are you saying does it this way?