A simple test that would gain you a Nobel prize

[Gizmotron]

The thing is that all interpretations of probability agree on the results, on the bottom line! Otherwise there would be different probabilities (percentages) according to different interpretations. Which is not the case.
The issue you raise is very important but theoretical/philosophical.

It's more pragmatic in fact. I know what you will find as the proof pretty soon. Assuming
that you intend to get to the bottom line, you will have passed one of the few very big
lessons in researching Roulette. This is your baptism of fire. Hopefully it will not kill off
your effort to find a great solution to beating this game.

[speed]

Interesting, i see Flatino is now expert for deviations 

[Turner]

Interesting, i see F_LAT_INO is now expert for deviations 

I see Flatino is a generous poster who shares his ideas

[speed]

 possible

[Robeenhuut]

Sam,

My findings in short (using the program): when in 50 spins a color hits 12 times (or less), in the next 50 spins, it has over 65% probability of being the dominant color with over 25 hits.

Kav

Lets say that you tested 10000 50 spins sessions where 65% ended with a less dominant color hitting more in following 50 spins. Do you think that you tested enough sessions to give you a statistically reliable breakdown and in real sessions you will be able to get this ratio? 

[F_LAT_INO]

I see F_LAT_INO is a generous poster who shares his ideas

--And I'm Croatian and this Speed chap is Serbian....that's why
these sarcastic remarks,Turner.But me don't bother,let him bark.

[Bayes]

Kav & RH,

So many theories are put forward on these forums but hardly any of them are adequately tested. My money is on the mathematical orthodoxy to prevail but it's fun to test the ideas so let's see what we get after 10,000 runs. Then we can abide by the result whatever it is, and move on to explore other ideas.

So here are my first set of results using the software. I encourage anyone who's downloaded it to post theirs too, that way we can get to the 10,000 session target more quickly.

The settings are: trigger = 12, 1st sample = 2nd sample = 50.

[attachimg=1]

So these results give a % of 508/924 = 55% 

Not looking so good now.

The % has to be significantly above this (at least 25 wins in 50 spins):

[attachimg=2]

[Gizmotron]

Kav -" Do we believe that past spins have no relevance or do we believe that (in certain situations) past spins can offer a guide for future results?"

This topic has been discussed for years. "Each spin is an independent trial."
But that doesn't prevent anyone from assigning meaning to sequences of
spins in order to identify typical occurring patterns or characteristics. At no
time does this form of assignment provide a source of prediction. It can
only provide as a source for confirmation, after the facts. Confirmation of
continuing events can serve as a bet selection method. At no time does it
serve as a predictor.

Now in your hypothesis if certain thing happens then something else will
happen in the following 50 spins. That's impossible. What force could cause
a thing to happen? Roulette is an independent game from spin to spin.

[TwoCatSam]

R.D. Ellison had a long a tiring discussion with three math professors--or was it two?  I believe it was two.  Anyway, he posed such questions to them that they had to admit there was some sort of memory, but they could not understand what it was.  Ellison called it the statistical "pressure" to conform to the norm....some such wording.

Anyway, I've studied Ellison's 3QA system enough to know there is something behind it.  Memory--whatever--there is something there.

Sam

Kav

I totally understand what you're saying.  Studied and thought much on it.  I'd love to see it come out of Bayes program--namely, the second fifty catches up somewhat with the first fifty........all the time every time. 

A guy based a system I have on this idea; the Magic Five, I think.  Says it makes a whopping profit.  He only uses five spins.

Sam

Edit:  What force, Giz asks.  There's the rub.

[Bayes]

Regression to the mean is a real phenomenon, but it only says that given an extreme event, the following event will be less extreme - the vast majority of the time. That's a long way from saying that events are  governed by an invisible elastic band which somehow pulls the outcomes back to balance in the short term. In fact the opposite is the case - RATIOS will tend to balance, but not absolute numbers, which actually diverge as you get more results.

[ginger]

Hello Kav,

This is what you wrote:

But the big thing is this:
Do we believe that past spins have no relevance or do we believe that (in certain situations) past spins can offer a guide for future results?
This is really big!

********************************************

As some members already know I play the dealers signature.
Each Dealer has his own Signature if they spin the Ball & Wheel , as soon as another Dealer takes over the system fauls.

So you have to wait for at least 10 spins as a "new" Dealer takes over , I also call it the Dealers KEY.

I'm honest to you that I know nothing of Math in Roulette , I concentrate on the numbers on the bord , take f.i. Dublinbet they show you the last 10 numbers that has been hit.

It is the  "TRICK" to find the right combi to make a good profit , the strange thing is that this key always work , but don't forget that each Dealer has his own key.

So if you ask yourself the question is there a memory in the past spins...I have to say yes..but a key sounds better to me.

Have a nice weekend


John      Rotterdam

[Drazen]

Mr. Bayes if you could please explain this thing for me, and say few words, or maybe coded something like this to check?

How could we expect to behave/or what distribution to get in 2nd group after our first deviation, but if 2nd is same as first?

Ok lets take that our famous example with 12 wins in 50 spins which is std over 3 already, and next 50 spins sequence we get same distribution again, which would be very strong deviation nicely over 4std (very very rare but still possible  knowing roulette deviations)?

We can be sure that in 3rd measured group we cant have such situation again because that would be std that isnt recorded in roulette history. But again is now sure that we will must have over 50% of wins?

How regression would behave now?

Maybe I asked this question clumsy, but I think you could assume what i wanted to ask?

Regards

Drazen

[TwoCatSam]

We can be sure that in 3rd measured group we can't have such situation again because that would be std that isnt recorded in roulette history. But again is now sure that we will must have over 50% of wins?

We can be sure we'd grow old (older) waiting for this phenomenon.

Sam

[Bayes]

Some more sessions and an update on the winning % :

[attachimg=1]

So pooling these results with my previous sessions, we have a total of 924 + 742 = 1666 sessions, and a total of 508 + 351 = 859 winning sessions, giving a current winning % of 859/1666 = 51.6%

[Bayes]

That is a very common misconception.
This is like saying that an imbalance would never be turned over; that if a color would establish a dominance it would keep it forever. Which is not the case.

To understand my way of thinking you can also read about the ergodic hypothesis.

Here is a link to Ellisons article about The Law of Statistical Propensity Sam is referencing.

Kav (Roulette probabilities)

Ok, your point about dominance is a fair one. It would be more accurate to say that imbalance in terms of R vs B doesn't systematically tend towards a limiting value, as the ratio does. What happens is that the difference between reds and blacks diverges and converges (in a cyclic fashion), but the number of "equalizations" is much much less than you would think. John Haigh covers this in his book "Taking Chances". There is even a formula which tells you the number of times you can expect red and black to balance in X spins, and the most likely scenario is 0 times.

Regarding Ellison's article, he seems to be trying to make the case that there's an inconsistency between statistical independence and the law of large numbers, but surely it's just common sense that the two aren't necessarily mutually exclusive. Yes, there is a "statistical propensity" for numbers to equalize, but this is just a consequence of the symmetry of the wheel. If there is symmetry in the device which generates the outcomes, then the obvious "solution" is simply that the symmetry determines that each number or group of numbers will show the same number of times (in the absence of any bias). The number 13 will hit as many times as any other number because it's the "nature" of 13 to tend towards 1/37, just like any other number, and what would be odd is if that didn't happen. So it seems to be sophistry on Ellison's part, and you admit yourself that the article was a plug to sell his book.