Quote:WizardI'm sure there are a lot of baccarat players would would agree with you. However, don't expect much support for your statement here.

People will say bac is different because you're dealing

down static decks. It is essentially the same as red and

black in roulette, however. Player banker outcomes are

quite random.

Quote:thecesspitPlease give me an example relating to a Roulette wheel, or other random number sequence.

By example, do you mean give a demonstration of

how you can look at past results and accurately guess

at a future outcome? I'm just discussing theory here,

demo's are another subject entirely..

Quote:EvenBobBy example, do you mean give a demonstration of

how you can look at past results and accurately guess

at a future outcome? I'm just discussing theory here,

demo's are another subject entirely..

Well, I see little theory here based on a stream of random numbers, so I thought seeing as I am not understanding your theory, I thought a practical example might help.

I'll state it again. -Either- the numbers from a Roulette wheel are random in the classic mathematical meaning OR they are not random by the definition.

Unless I've made a mistake. If I have, fine, point it out.

Quote:thecesspitThe Law of Large numbers is EMERGENT from the basic probability of the game. The chicken and egg is here solved.... the low level behaviour came first.

There's no proof of that. It looks like they emerged

at the same time, by Cardano in the middle of the

16th century. Bernoulli later proved it. In any event,

LLN is now the god of probability theory, not the

other way around.

Quote:thecesspitI'll state it again. -Either- the numbers from a Roulette wheel are random in the classic mathematical meaning

Of course they're random, everything hinges on them being

pure random. Thats why I won't use numbers spewed by a

computer RNG, they're pseudo random. In the long term

they look the same as pure random. In the short term, where

we use them, they're flawed, if ever so slightly.

Its funny, I use a tool called RX Roulette when I practice,

Sometimes I forget to use imported actual spins for a session

and inadvertantly use the RNG they provide. After awhile I

start to get pissed because I'm making so many mistakes,

and then I realize its the damned RNG I'm using and not

the real thing. Pretty funny.

Quote:EvenBobThere's no proof of that. It looks like they emerged

at the same time, by Cardano in the middle of the

16th century. Bernoulli later proved it. In any event,

LLN is now the god of probability theory, not the

other way around.

You misunderstand what I mean.

Not which idea came first, but that one (basic probability systems) can prove the other. The Law of Large Numbers emerges from the axioms of probability maths. You can prove the LLN from the axioms of probability. I am not sure you can prove probability mathematics based on the law of large numbers as an axiom.

It may be one of the more important laws in probability maths, but doesn't mean it's not emergent phenomena from the base.

Quote:EvenBobOf course they're random, everything hinges on them being

pure random. Thats why I won't use numbers spewed by a

computer RNG, they're pseudo random. In the long term

they look the same as pure random. In the short term, where

we use them, they're flawed, if ever so slightly.

Its funny, I use a tool called RX Roulette when I practice,

Sometimes I forget to use imported actual spins for a session

and inadvertantly use the RNG they provide. After awhile I

start to get pissed because I'm making so many mistakes,

and then I realize its the damned RNG I'm using and not

the real thing. Pretty funny.

They cannot be pure random and show the effect you are theorizing about. You need to point out to me where my argument is flawed. A true stream of random numbers must be independent of one another, or they ARE NOT RANDOM. It's really that simple. If you can see a short term effect, THEY ARE NOT PURE RANDOM numbers. That is my statement, my core argument. Without evidence/argument to the contrary, this whole thing is pointless.

Isn't it more likely that there's actually a subtle effect in the real roulette numbers that makes them less random that you don't see in a Random number generator?

Quote:thecesspitthe axioms of probability maths.

Probability 'theory' is a squishy subject. Probability

math kinda sorta works, but there are no probability

laws, per se. They seem to invent a new offshoot

of probability every other year. One of the more

recent ones, Imprecise Probability, didn't get named

till the 90's.

IP

Quote:thecesspitA true stream of random numbers must be independent of one another, or they ARE NOT RANDOM.

They are completely independent of each other, thats

why they're so reliable to use. Thats why probability

works on them, because of their complete indepencence.

Computer RNG's produce non independent outcomes too

often, they are uneliable. They're fine for something crude

like a slot machine, but they're flawed for accurate

probability work. An example here would only be confusing,

it would be meaningless. The outcomes have to be true

random or they won't work.

Quote:EvenBobProbability 'theory' is a squishy subject. Probability

math kinda sorta works, but there are no probability

laws, per se. They seem to invent a new offshoot

of probability every other year. One of the more

recent ones, Imprecise Probability, didn't get named

till the 90's.

IP

Did you actually read the article? Did you understand what it is talking about? Or just seeing the word "probability" in the title was enough for you?

I am not sure what you refer to as "laws", it is more common to use terminology of axioms and theorems. Like any non-trivial and self-consistent axiomatic theory, probability contains an infinite number of theorems, provable on a finite axiom base. Law of large numbers is but one of those theorems.

There is nothing "squishy" about probability theory. It is a closed, self-consistent formal theory, much like calculus, algebra or games theory. Among other things, much if not most of contemporary physics could not be formulated without probability.

Quote:weaselmanDid you actually read the article? >>There is nothing "squishy" about probability theory.

Read it half a dozen times in the last week. And

PT is squishy. Thats why probabilists are always

fighting with each other.

"It is unanimously agreed that statistics depends on probability. But, as to what probability is and how it is connected with statistics, there has seldom been such complete disagreement and breakdown of communication since the Tower of Babel." Wikipedia.

Quote:EvenBob

"It is unanimously agreed that statistics depends on probability. But, as to what probability is and how it is connected with statistics, there has seldom been such complete disagreement and breakdown of communication since the Tower of Babel." Wikipedia.

I'll add the last sentence of the quote, you omitted, I am sure, accidentally ...

Doubtless, much of the disagreement is merely terminological and would disappear under sufficiently sharp analysis.

This particular phrase belongs, I think, to Leonard Savage, who was more of an economist than a mathematician. He wrote a book "Foundation of Statistics", where he suggested what he called "subjective probability" (or, perhaps, "personal probability" - I don't remember exactly) as a substitute for the formal measure-theoretical interpretation.

The disagreement he refers to in this statement is the terminological disagreement between mathematicians, statisticians, economists, engineers and philosophers, who use a number of identical terms (probability, frequency, objective, physical etc.) to mean quite different things.

As long as we talk about mathematical theory of probability and mathematical statistics, none of this applies at all. There is nothing "squishy" or ambiguous about those theories.

Quote:weaselmanit does not make it any more believable or give it any more weight than your (quite weightless lately) own posts.

Uh Huh, if you can't dispute the message, kill

the messenger. In this case, Wiki. I happen to

read a lot of probability articles, and I see constant

bickering and disagreement. Don't you find that

odd for a 'science' that you say is so precise and

accurate? Its squishy, thats the perfect word to

describe it.

Quote:EvenBobUh Huh, if you can't dispute the message, kill

the messenger. In this case Wiki

I am not disputing the message, just your interpretation of it.

The message was incomplete (I just added the missing sentence above in my earlier post, probably, after you posted yours), which, probably has contributed to your wrong interpretation.

Nothing is wrong with wiki as it turns out, just your inability to cut and paste properly (unless, you actually truncated the quote on purpose, which I don't want to believe).

Quote:I happen to

read a lot of probability articles, and I see constant

bickering and disagreement. Don't you find that

odd for a 'science' that you say is so precise and

accurate?

No, not at all. The same is going on with cosmology, special relativity, genetics, paleo-biology etc., etc., even the Super-String theory.

What's common between those fields, is that their are very highly popularized, and described at "layman level", making them look simple enough for a layman to actually believe that he understands. It takes one sentence, by somebody, not very competent in the field, or even, by someone competent enough, but not very good at written language, to fuel months if not years of "bickering and disagreements" in the "popular-science" circles.

Did you notice by any chance how much fewer "bickering" and "disagreement" is out there about, say, General Relativity, which is totally and completely based on its Special cousin?

Or, back to the probabilities, would you say you are noticing more disagreement in the discreet case (combinatorics, binomial trials etc.) or in the continuum or transfinite applications?

Quote:weaselmanI can dispute the message.

It doesn't matter, whats relevant to me is, no math,

no probability, governs the next outcome of

random numbers. All you know is what will 'probably'

happen 'eventually'. Up until that eventuality, you

can drive a truck thru the lack of knowledge about the

next spin, its that imprecise.

Quote:EvenBobIt doesn't matter, whats relevant to me is, no math,

no probability, governs the next outcome of

random numbers.

This is actually true. It is physics (not the scientific field, but the laws of nature) that governs the outcome in case of roulette. Math just tells us quantitatively which outcomes are more or less likely than others. But so what?

Perhaps, it's time to end the suspense and reveal the answer to the question you started your thread with - where are you going with this?

Quote:weaselmanThis is actually true. It is physics (not the scientific field, but the laws of nature) that governs the outcome in case of roulette.

Thats why its important that the wheel is unbiased

and produces true random results.

Quote:EvenBobThats why its important that the wheel is unbiased and produces true random results.

Well, yeah ... In general, given a game with some stated rules, it is important that it is played according to those rules.

But that's not "why". In fact, I don't see anything whatsoever this statement has to do specifically with either roulette or statistics or with anything at all mentioned in this thread earlier.

Quote:weaselmanI don't see anything whatsoever this statement has to do specifically with either roulette or statistics

Only truly independent outcomes are accurate enough

to work with.

Quote:EvenBobOnly truly independent outcomes are accurate enough

to work with.

Truly independent outcomes contain no information(*). If they do, they are not truly independent.

Here's a thought experiment :

Say you have a way of looking at a stream of random data from a wheel and making better than average guess at the next spin, so that you can beat the house edge.

Randomness means I should be able to rearrange that stream of data (shuffle it) and there is no difference. In fact, I should be able to take every fifith sample and present it. In fact, I could take (truly) random samples from the stream and present it as a new stream. None of those streams are any different in the data type if they are random.

Would the method still be valid? If the method exists, and the data is truly random, it should do. I've changed nothing in terms of the data.

(*) I'm not 100% sure that is true. The information they may contain is that they look random... and that's information...

Quote:thecesspitWould the method still be valid? If the method exists, and the data is truly random, it should do. I've changed nothing in terms of the data.

Its valid and I've proved it. I've taken spins from

3 different wheels at the same time. 1 from wheel

A, the next one from wheel B, and so on, till I have

a string of random outcomes. The results are the

same as if I took them from a single wheel.

So you don't care even what order the random stream comes to you either (as per my example) or which data you are given from where?

Does that mean you could get a random stream from wheel A and use that for wheel C?

it. It worked OK, not as good as using the

actuals from just one wheel. It was kind of

spooky, actually. But it had to work, or I

would have really been at a loss for an

explanation.

Quote:TheNightflyOk, let's say the bag has 38 balls of which 20 are red and 18 are blue. The guy betting on red still has over a 5% advantage. What's your point? There's no way of knowing what ball he'll pull but you'd have to be a pretty big moron to bet on blue.

How about assuming 18 black balls, 18 red balls, and 2 green balls? Then you'd really be a moron to bet on blue. (LOL)

Quote:EvenBobOnly truly independent outcomes are accurate enough

to work with.

This is a short unqualified statement, and, like most such statements, it is completely meaningless.

To actually convey some meaning, you would have to qualify what you mean by "truly", by "independent", and by "accurate", and also explain what kind of "work" you are planning on doing with these outcomes, and what kind of result you expect to get.

For example, if you are using the word "independent" in the common statistical sense, requiring the random outcomes to be independent on each other, by "work" you mean performing some kind of (presumably, arithmetical) operations on the numbers, representing past outcomes, and the result you are expecting is being able to predict future outcomes with better than expected frequency, then your original statement is false.

Note, that it may be possible to predict future outcomes by analyzing past history whether they are "truly independent" (e.g., biased wheel), or not (e.g. red always hits after black). So, as long as by "work" you mean "devising a method of predicting future results", the ability to do that work has absolutely nothing to do with results being independent (of each other).

Not also, that, contrary to common belief, modern software RNGs, with an exception of a few simplest ones, actually produce random numbers, that are statistically independent of each other. There are various sources of entropy that can be used to achieve that - such as hard drive movements, network noise, cpu clock and temperature etc.

Quote:EvenBobBut it had to work, or I

would have really been at a loss for an

explanation.

Are you saying that you are not at a loss as it is? In other words, you seem to be suggesting that you actually have an explanation for your "method". If so, that's just great! What are you waiting for? Let's hear it!

Quote:EvenBobThere are no laws or theories that apply to short term play. The casino can

lump all the players together and get a long term number. On a player

to player basis, however, they're lost. Nothing can apply to the next bet

you make in a game of random outcomes.

In case it hasn't yet been pointed out, this is incorrect. The math for short-term and long-term results in a game of independent trials is exactly the same. The multinomial distribution describes games like roulette and it holds for any value of n (number of trials).

Like Jerry Logan and Evenbob, John Patrick is also a good example.

Quote:weaselmanqualify what you mean by "truly", by "independent",

Coming from an unbiased real wheel.

Quote:MathExtremistThe math for short-term and long-term results in a game of independent trials is exactly the same.

So what will the next spin be, red or black? I can

figure out what the approximate distribution will

be for the next 10,000 but they won't let me bet

on that. Whats the math that accurately predicts

the next spin.

You CAN bet on the next 10,000 spins and the approximate distribution if you so wish. Just means you gotta stand there for a LOOOONG time.

Quote:EvenBobComing from an unbiased real wheel.

You are wrong.

Quote:weaselmanYou are wrong.

You don't think an unbiased wheel produces

independent outcomes?

Quote:EvenBobYou don't think an unbiased wheel produces

independent outcomes?

I do. I also think, that a biased wheel produces independent outcomes.

But you still have not explained what it is you want to do with those outcomes to begin with, so, I don't really know what it has to do with anything.

Quote:EvenBobSo what will the next spin be, red or black? I can

figure out what the approximate distribution will

be for the next 10,000 but they won't let me bet

on that. Whats the math that accurately predicts

the next spin.

Yet quite counter-intuitively, your alleged ability to figure out the approximate distribution of the next 10,000 outcomes will actually be less accurate, on an absolute basis, than your prediction of the next single outcome. How many reds will appear in the next 10,000 outcomes? The expected number is 4737, yet the chances of there being between 4736 to 4738 reds is very low. In a single spin, the chances of there being 0 or 1 reds is 100%.

Richard A. Epstein, in "The Theory of Gambling and Statistical Logic", puts it quite elegantly:

"The law of large numbers has frequently been cited as the guarantor of an eventual head-tail balance. Actually, in colloquial form, the law proclaims that the difference between the number of heads and the number of tails thrown may be expected to increase indefinitely as the number of trials increases, although by decreasing proportions." (p. 28, emphasis mine)

And the casino will absolutely let you bet on 10,000 spins. Since all of them are independent, it doesn't matter whether you bet sequentially or simultaneously. And again, due to independence, the casino doesn't need to change its per-spin payout odds.

As before, the "math that accurately predicts the next spin" is the same math that is used to describe the aggregation of results. If you're looking at a specific bet and its chances of winning then the problem decomposes into a Bernoulli trial. For "red", p(win)=18/38.

Here's the real question. Since it seems that 18/38 isn't accurate enough for you for a single trial, why should it be an acceptable figure to use as the basis for your 10,000 trial analysis?

Quote:MathExtremist

And the casino will absolutely let you bet on 10,000 spins.

Good grief. I'd forgotten what its like having discussions with

math people. They take everything literally and always miss

the nuance.

Never mind...

Quote:EvenBobGood grief. I'd forgotten what its like having discussions with

math people. They take everything literally and always miss

the nuance.

Never mind...

You are the only person here, math-orientated or otherwise, who perceives any nuance at all in the question...

Perahps you should try the John Patrick forum.

On that forum they don't worry about the facts, the math, or proof. They instead, believe in following feelings, nuance, hunches, folklore, superstition, and mysticism.

Quote:EvenBobGood grief. I'd forgotten what its like having discussions with

math people. They take everything literally and always miss

the nuance.

Never mind...

The nuance that's missing here is your understanding of variance. The only difference between a single random trial and a lot of independent random trials is the variance for a large aggregation is smaller, relatively speaking, than for a single outcome. The issue you need to resolve is that you *think*, for whatever reason, that there should be some way to predict short-term results because the long-term results seem predictable to you. That's your error, because the same math that describes the unpredictability of the short term is what also describes the relative predictability of the longer term. If it didn't, lots of things would be broken.

As you never go for independent testing (*), I wonder why you waste your time starting down these conversational routes. They always end up the same place : you claim a method that works for you, everyone says what you claim is not possible, and you say "well it is, I've tested it, it works, believe me".

(*) I understand the logic you use behind that.

(**) You can't any more, they've taken the money for other work in the skeptical community. But maybe someone else out there is offering such a prize.

it was about psychic ability. He would never let

somebody stand by a wheel, writing down the

results, thats against everything he stands for.

Quote:EvenBobThe Randi challenge was not about probability,

it was about psychic ability. He would never let

somebody stand by a wheel, writing down the

results, thats against everything he stands for.

I'm betting he's smart enough to know that it wouldn't help you...

"Webster’s Online Dictionary defines “paranormal” as “not scientifically explainable; supernatural.”

http://www.randi.org/site/index.php/1m-challenge.html

Quote:rdw4potusI'm betting he's smart enough to know that it wouldn't help you...

Sigh. I've had this stupid conversation a dozen times

in the last 5 years. Randi is NOT looking for AP casino

players! He's looking for paranormal activity, mind readers,

fortune tellers. He could care less about card counting

or how to beat any casino game legitimately.

Quote:EvenBobHe's looking for paranormal activity, mind readers,

fortune tellers. He could care less about card counting

or how to beat any casino game legitimately.

I am sure you are mistaken, try contacting him, he'll be very interested.

As it has been pointed out earlier, "paranormal" means "unexplained by science", which is exactly what your method is.

Just contact him, and explain that you are able to predict outcomes of roulette wheel spins based on the past history (from a different wheel(!)). He'll be all over you!

Quote:weaselmanwhich is exactly what your method is

What is my method, exactly. Oh yeah, probability. You

think Randi wants AP casino folks bothering him, go

right ahead. You don't have a clue.

Quote:EvenBobYou

think Randi wants AP casino folks bothering him, go

right ahead.

No, I don't think he wants that. I also do not think that you are that.

I think, he'll love you. Why? See previous post.

Quote:weaselmanNo, I don't think he wants that. I also do not think that you are that.

I think, he'll love you. Why? See previous post.

Like I said, I've dealt with the shear stupidity of

this for years. He's not going to test people who

beat casino games by using probability. Why in

the hell would he.

Quote:EvenBobHe's not going to test people who

beat casino games by using probability. Why in

the hell would he.

Because it is paranormal. That's what he does.