Do RNGs produce random outcomes?

[ThodorisK]

(a counter-argument besides REASON No3, lol)

 

[aka23]

The National Institute of Standards of Technology's test suite for RNGs is summarized at Old / Expired Link and described in detail elsewhere on the site. Note that they don't focus on local and global randomness, like the Wikipedia introduction to "statistical randomness" that you read. Nor do they imply that RNGs would be rejected due to "local" streaks of the type you have mentioned.

 

[ThodorisK]

All major sofwares use these tests? if yes, how do you know this?

 

[spearmaster]

I'm still having trouble trying to figure out what you are trying to prove - maybe it has something to do with the fact that I only read your posts in the morning or something.

If I guess right, you are saying that RNGs can't be random because it is possible to generate a series of numbers (0-1) which would fail a test of randomness?

Somehow that doesn't make a whole lot of sense...

 

[aka23]

All major sofwares use these tests? if yes, how do you know this?

I expect P-RNGs used by most casino softwares are based on one of a few long tried and tested algorithms. For example, iTechLabs, which tests some casino softwares, recommends the 5 following algorithms on the page at Old / Expired Link :

1. Mersenne Twister by Matsumoto and Nishimura (MT19937).
2. RANLUX by Luscher.
3. L'Ecuyer's Combined Multiple Recursive Generator.
4. L'Ecuyer's Tausworthe generator.
5. Ziff's Implementation of the gfsr4 generator.

If you search for any of these names, you'll find they have passed the NIST tests describe above.

After the RNG is in place, many of the larger casino softwares hire an outside auditor to confirm their RNG. I suspect this is primarily done by reviewing game results over a long period. An example, is the summary statement PWC game for Microgaming gaming roulette over a long period, as described at Outdated URL (Invalid) .

 

[ThodorisK]

To make it more simple, after you see e.g. 20 pictures in a row, the probability for the next card to be a picture is still 4/13? Doesnt the fact that the RNGs have passed certain statistical tests, lessen the probability below 4/13?

One might argue that a sequence made out of really random-independent outomes, would certainly (or almost certainly) pass with success the same statistical tests that the RNGs pass with success.
And therefore, the successful passing of some stat tests, does not create a bias.

But the latter counter-argument gives us the next counter-argument:
Suppose that a real roulette gives outomes which are random-independent between them:
It's another thing to know that some PAST results of a real roulette have successfully passed some stat tests, and another thing TO KNOW WITH A 100% CERTAINTY that ALL THE FUTURE results of this real roulette will successfully pass these tests. Because in this case, the results of this real roulette are not random-independent, because if they were, the probability that the future results would pass the stat tests successfully, would not be 100%, but it would be e.g. 99%!
Are we 100% or 99% certain that a RNG which has passed some stat tests in the past, will pass them again in the future? I think we are 100% and not 99%. Why? I am still thinking on this why. Perhaps I have already prooved it in my previous posts. Or, perhaps I am wrong. Can you help me on this? Anybody?

 

[pokeraddict]

Thodo, have you stopped gambling yet?

 

[spearmaster]

It's another thing to know that some PAST results of a real roulette have successfully passed some stat tests, and another thing TO KNOW WITH A 100% CERTAINTY that ALL THE FUTURE results of this real roulette will successfully pass these tests. Because in this case, the results of this real roulette are not random-independent, because if they were, the probability that the future results would pass the stat tests successfully, would not be 100%, but it would be e.g. 99%!

Pardon me for saying this - in a past post, you did say that your English wasn't that good at the time you were studying - but it appears to me your English is perfectly good - in fact, too good

But instead of mathematical/scientific/geekspeak, which unfortunately I can't be bothered to decipher if it isn't important, would you mind stating your theory in plain English?

I have enough trouble dealing with contracts and stuff like that and I would prefer that my leisure time only be spent enjoying reading and responding to posts... rather than trying to figure out exactly what you are getting at.

Thanks

 

[aka23]

When a casino shuffles, the cards are arranged in a random order using the RNG. The casino then deals cards in the selected order. The chance of drawing a picture card at any given time is (# of picture cards left in decks /# of cards in decks). This probability is 4/13 on the first card dealt after the deck is shuffled. The probability is slightly more or less than 4/13 on the 2nd card dealt, as the fraction (# of picture cards left in decks /# of cards in decks) changes.

Statistical tests rarely return results with a 100% confidence level. I believe most groups that audit casinos uses a 95.0% confidence level for this testing.

 

[ThodorisK]

Forget about the 4/14 etc. Focus on this:

I think the confidence level has nothing to do with what I meant. A confidence level refers to the stat tests made to outcomes that have happened in the past. Am I wrong? Explain clearly.

What I meant is that because the RNG outcomes are produced in a particular way e.g. a logarithm and/or the motion of molecules, electrons, etc, (and perhaps, also because they the FUTURE outcomes are FORCED to obey to some statistical tests)
then:
(perhaps, I cannot be sure),
there are many cases that there is a 100% probability that the FUTURE outcomes of a RNG will succeed these stat tests.
And if the results were truly random-independent, this 100% would not be the case.

This could have as a consequence, either that a player can take an advantage of this bias (as I explain in the REASON No1 paragraph), or it could mean that the average payoff is pretedermined to happen (REASON No3). I dont want to analyse these complications yet.

Do you know any other forum (or anything like a forum) where I can discuss this matter with other mathematicians?

 

[aka23]

RNG tests don't simply report pass/fail. They report the probability of the test results occurring in a random distribution. The confidence level relates to how far off from expected values this probability can be for the test to pass. We arbitrarily designate a certain probability as passing. There is rarely a 100% probability of such a test passing. The probability of a test passing relates to the specified confidence level, and can be calculated. Some RNG algorithms have a higher failure rate than others. The page at

You do not have permission to view link Log in or register now.

lists the failure rate for some RNGs on the NIST test I mentioned earlier. The author of the page is saying his RNG is good because it has a failure rate of 0.175%.

If you want to find a forum with mathematicians, you can probably do so with Google. Note that a couple posters on this forum are mathematicians.

 

[ThodorisK]

So, these confidence levels are refering to the PAST outcomes of the RNG.
So ... I dont want to repeat what I said previously.

I guess the only one who can help me judging my arguments, is myself.
But I dont have enough knowledge on the exact procedure they use for the generation of the RNG outcomes. Do you completelly understand that Boss media chart? Will Moss Media give more more details on their procudure?

 

[ThodorisK]

It truly isn't possible for a computer to come up with a coincidental account of numbers. A computer always calculates its numbers based on mathematical statistics. These statistics remain constant regardless of the complexity of the calculating process.........Boss Medias RNG is tested over billions of results that are constantly monitored for accuracy in our 100% random results. You can follow up your interest by looking at the following flow chart in Picture 14: Flowchart
( Link Removed ( Old/Invalid) )

I still see that what is meant by this guy's words, is that the future outcomes (sequences of digits), before they come out, are filtered by stat tests, so that these outcomes will succeed with a 100% certainty in these tests.

Perhaps he mis-formulated what he meant? I dont think sooo!

 

[aka23]

So, these confidence levels are refering to the PAST outcomes of the RNG.
So ... I dont want to repeat what I said previously.

I guess the only one who can help me judging my arguments, is myself.
But I dont have enough knowledge on the exact procedure they use for the generation of the RNG outcomes. Do you completelly understand that Boss media chart? Will Moss Media give more more details on their procudure?

A RNG is tested with a long sequence of outputs. The result of the RNG test does not use past outcomes of the RNG, only the sequence under test. The outcomes of the RNG test is compared to the the expected outcome in a true random distribution. The confidence interval relates to the distance between the results and the expected results in a true random distribution. These expected results are determined using probability equations.

The Boss Media guy said that their results are "monitored" for accuracy, not that the results are filtered before they come out. They may simply mean that results are audited by a 3rd party, like the other large softwares do. Filtering results before they come out would defeat the purpose of a RNG. And yes, I do understand the basics of the Boss Media flowchart. I'd need more detail about the procedure to "understand completely".

The problem is not that nobody on this forum is able to judge your arguments. The problem is that you do not like the result.

 

[ThodorisK]

By this "100% certaintly" I simply mean that the future outcomes are bound to succeed these tests, simply because they have already been filtred by them before they come out.

It has nothing to do with the "degree of certainty" or else called "confidece level" or "signifocance level" of a hypothesis tests.

Yes, I know, Nolan, you say that's not what he meant. But that's exactly what he expressed.

I also said that even if this is not the case, then for many other reasons, perhaps in many cases the probability of the FUTURE outomes to succeed many stat tests, is 100%, whereas if they were truly random, this probability would be less than 100%.

 

[ThodorisK]

Ok, then, since you understand the chart of Boss Media.

The first data comes from the "noice" which is the motion of molecules or subatimic particles, right?

Then? This data is entered into a logarithm (a mathematical function-equetion), or not?

 

[ThodorisK]

And explain the procedures of the
Picture 14: Flowchart
which is the page:
Link Removed ( Old/Invalid)

If I ask Boss Media to explain this chart to me, will they do it?

 

[tennis_balls]

And explain the procedures of the
Picture 14: Flowchart
which is the page:
Link Removed ( Old/Invalid)

If I ask Boss Media to explain this chart to me, will they do it?

you go Thodo!

i think you've really got him pinned to the mat now!

CHECK! and MATE!

finally this matter is settled!

 

[aka23]

The Orion RNG uses two Zener diode noise sources to generate "random" numbers. This Orion RNG feeds reseeding, key, and other data into a Mersenne Twister P-RNG with Blowfish cipher encryption that has several different pipes. The 32-bit output of this Mersenne Twister/Blowfish P-RNG is sent to a "random number client."

 

[ThodorisK]

"... there are sites using rogue software that must be exposed ...
...The software is very good, with great control over the profit/winnings. It has the ability to change the winning % per game/table in real-time. Also you can withdraw the profit and limit player's winnings. Games are completely random, if the winning is within the limit. If the bet is causing winning over the set limit, the generator is generating another number until the winning is within the limit (or player loses).
..."
Michael Shackleford (

You do not have permission to view link Log in or register now.

)

Compare the above meaning of these words, with the meaning of the bolow words:

"Sren: How do you make random numbers and a fair casino game?
Roger: It truly isn't possible for a computer to come up with a coincidental account of numbers. A computer always calculates its numbers based on mathematical statistics. These statistics remain constant regardless of the complexity of the calculating process. I have been responsible for the development of Boss Medias Random Number Generator 'RNG' ..."
Roger Jnsson (Link Removed ( Old/Invalid) )

I think Mr Roger has to re-formulate what he meant.

The 2 above meanings are the same: REASON No3:
"Are the future random outcomes cut and modified so that the profits/losses of each player, or the profits/losses of all players playing in a Boss Media casino, will definitely fall close to the statistically expected average payoff? Doesnt that mean that each player NO MATTER WHAT HIS DECISIONS ARE, will surely not win much above (or lose much below) the statistically expected average payoff? If the amount I will lose or win is predetermined to fall within certain limits, then not only the future outcomes are not random-independent events, but this equals to CHEATING. For example, if you are already winning much above the statistical average payoff, and you suddendly make a very big bet, this bet is bound to lose!!!"
Theodoros Kiriakopoulos

And of course, this cheating is impossible to be statistically prooven, as the results will be very close to the statistically average loss!

I honestly did not want to proove cheating when I started writing in this thread. I wanted to reveal and discuss my possible discovery on how to make an edge for the player out of the degree of the non-randomness (REASON No1). But the day this thread started, I found Roger's words! Shit!

But I still think that there might be a way to translate the degree of non-randomness into an edge for the player, if Roger meant something else, and if most casinos dont apply REASON No3 (as Aka23 is 100% certain this is the case).

So I dont want to discuss again cheating in this thread and REASON No3. I think I have exhausted the matter.

I only want to discuss if it is possible to take advantage the degree of non-randomness to create an edge for the player (supposing there is no cheating, not even in the REASON No3 form of cheating). So supposing this, I start again about if it is possible for the player to have an edge:

 

[spearmaster]

For example, if you are already winning much above the statistical average payoff, and you suddendly make a very big bet, this bet is bound to lose!!!"

This is what is known as "gambler's fallacy".

You do not have permission to view link Log in or register now.

Please understand this: There is no such thing as an event being "due." An event is not more likely just because it has not happened for a long time. For example, many people mistakenly believe that if one color in roulette has won several times in a row then the other color is overdue and they should bet on it. While the ratio of reds to blacks will always approach 50/50 in the long term, it can not be concluded that this will happen in the short term. It does not matter what the history of past spins is; every trial in games of luck like roulette are independent, and each color is equally likely to come up every time. If you don't believe me, try guessing the toss of 1,000 coin flips using any method of prediction you like. I guarantee that 99% of the time you will get between 459 and 541 correct.

The idea that events can be "due" is known as the Gambler's Fallacy.

 

[KasinoKing]

Sheeesh this is one of the most boring threads I ever tried to read!

I echo Spear:-

But instead of mathematical/scientific/geekspeak, which unfortunately I can't be bothered to decipher if it isn't important, would you mind stating your theory in plain English?

So are you saying casino software is rigged, RNG's are not 100% truly random (which no one can detect & doesn't matter a toss to the person playing the game anyway), it's impossible to beat the house edge, or maybe something which we have not already heard 1000+ times before...?
(That question is to Thodorisk, not Spear!)

This is what is known as "gambler's fallacy".

Originally Posted by Wizard of Odds

You do not have permission to view link Log in or register now.

Please understand this: There is no such thing as an event being "due." An event is not more likely just because it has not happened for a long time. For example, many people mistakenly believe that if one color in roulette has won several times in a row then the other color is overdue and they should bet on it. While the ratio of reds to blacks will always approach 50/50 in the long term, it can not be concluded that this will happen in the short term.

In my mind those two bold bits totally contradict each other.
It's OK - please no-one try to convince me otherwise!
I am quite happy to carry on winning in my ignorance and using my gamblers fallacy...

 

[tennis_balls]

I am quite happy to carry on winning in my ignorance and using my gamblers fallacy...

That's because you are on the top of a double-secret list that online casinos use to ensure that there are a few perrenial winners. The software recognizes KK's IP address and then re-directs his action to the +EV RNG.

 

[aka23]

I believe the Wizard of Odds quote relates to Start Your Own Casino software. This small software has advertised itself to be unfair to potential buyers. For a more detailed summary, see the rogue section of this site at https://www.casinomeister.com/casino-reviews/start-your-own/ . Start Your Own Casino software is (or has been) unfair, but this does not mean Boss Media or any other software has similar problems. The Boss Media rep's statement is poorly worded, yet clearly has a different meaning. While we are quoting Wizard of Odds, note the following statement about Boss Media BJ fairness in Wizard's Q/A column:

You need to give me some hard numbers to have this taken this seriously. For example if you played 1000 hands you would expect the dealer to have a 10 or face card up about 308 times. The probability of the actual number being within 50 of 308 is 99.93%. If you were outside of 50 then we could raise our eyebrows and if you were outside even more we could really get serious. However I can't do much with "the majority." I indicate how to gather data a test for online cheating in my FAQ. Finally, I want to say that I strongly feel that Boss Media is playing a fair game.

 

[ThodorisK]

Aka, do you know how often the "noise" re-seeds the logarithm?