Play roulette at Royal Panda

Author Topic: A Test for Randomness  (Read 2150 times)

Bayes

  • Moderator
  • Veteran Member
  • *****
  • Posts: 577
  • Thanked: 434 times
  • roulettician.com
A Test for Randomness
« on: May 24, 2016, 08:59:36 AM »
New article :  A Test for Randomness

The creative system designer will see some uses for this, I'm sure.  ;)


 
The following users thanked this post: kav, Reyth, MrPerfect.

kav

  • www.Roulette30.com
  • Administrator
  • Hero Member
  • *****
  • Posts: 1531
  • Thanked: 581 times
  • Gender: Male
Re: A Test for Randomness
« Reply #1 on: May 27, 2016, 09:12:03 PM »
Bayes,

This is a great topic.
I admit I haven't read your article in full yet.
But here are some issues that always bothers me, when I think about the issue of randomness and roulette.

Btw, I agree with you that physics is 'theoretical' in comparison with medicine and agriculture. But it is 'empirical' in comparison with mathematics. See more on this here.

Now to the main issue of randomness.
I see some relation between gambler's fallacy and randomness testing. I see some sort of logical and philosophical contradiction. Here are my thoughts.

If we believe that the result of each spin is totally independent, then every possible spin sequence is equally probable.
This means that this sequence: 5,5,5,5,5,5,5,6,6,6,6,6,7,7,7,7,7,7,8,8,8,8,8,......
is equally probable with this sequence: 36,30,1,1,4,9,17,17,4,30,30,36,18,22,6,7.....

Then on what grounds is it possible to evaluate a spin sequence and give a verdict on its randomness?

To ask it in another way. Is randomness something unknown or known? If it is totally unknown how can you evaluate it?
Does randomness has limits or not? If it doesn't then how can you evaluate it?

PS: I see a similar contradiction on the house edge argument . They say every spin is independent and unpredictable and nothing is due, yet they fully expect the 2,7% zero effect to stop you from winning, because it is due... But this is a totally different discussion and there is no point in arguing about this here.
 
The following users thanked this post: Reyth

Sheridan44

  • Mature Member
  • ***
  • Posts: 226
  • Thanked: 185 times
  • Gender: Male
Re: A Test for Randomness
« Reply #2 on: May 27, 2016, 09:47:01 PM »
 

Jesper

  • Hero Member
  • ******
  • Posts: 1138
  • Thanked: 535 times
  • Gender: Male
Re: A Test for Randomness
« Reply #3 on: May 28, 2016, 12:22:38 AM »
"This means that this sequence: 5,5,5,5,5,5,5,6,6,6,6,6,7,7,7,7,7,7,8,8,8,8,8,......is equally probable with this sequence: 36,30,1,1,4,9,17,17,4,30,30,36,18,22,6,7.....",

It sounds like it may not be, but it is the same probability.  We will see the first very seldom if ever, and the second as well.   Seven in a row I saw once, but never followed by a streak of six. The other sequence I have not noted, but it is possible I never got it in my personal play.

What makes the first "rare" is it is easy to recognise, and the play has rules we will win the same if we
bet them all and got it right in both sequence.
 
The following users thanked this post: kav

kav

  • www.Roulette30.com
  • Administrator
  • Hero Member
  • *****
  • Posts: 1531
  • Thanked: 581 times
  • Gender: Male
Re: A Test for Randomness
« Reply #4 on: May 28, 2016, 12:31:43 AM »
"This means that this sequence: 5,5,5,5,5,5,5,6,6,6,6,6,7,7,7,7,7,7,8,8,8,8,8,......is equally probable with this sequence: 36,30,1,1,4,9,17,17,4,30,30,36,18,22,6,7.....",

It sounds like it may not be, but it is the same probability.  We will see the first very seldom if ever, and the second as well.   Seven in a row I saw once, but never followed by a streak of six. The other sequence I have not noted, but it is possible I never got it in my personal play.

What makes the first "rare" is it is easy to recognize, and the play has rules we will win the same if we
bet them all and got it right in both sequence.

Agreed. But if "anything is possible" then you can not tell if something is truly random or not. Even a  5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5..... sequence falls within the limits of randomness. Because if one starts telling "this is random and this is not" and starts to exclude possible sequences, it means that random is not completely random.

To me "testing randomness" seems a contradiction in terms.
« Last Edit: May 28, 2016, 12:35:30 AM by kav »
 

Jesper

  • Hero Member
  • ******
  • Posts: 1138
  • Thanked: 535 times
  • Gender: Male
Re: A Test for Randomness
« Reply #5 on: May 28, 2016, 12:35:23 AM »
Test for randomness in such a small sample is not possible.  The randomness test are like they can tell a very high or low probability the sample is random, it can never say it is absolute sure random.
 

kav

  • www.Roulette30.com
  • Administrator
  • Hero Member
  • *****
  • Posts: 1531
  • Thanked: 581 times
  • Gender: Male
Re: A Test for Randomness
« Reply #6 on: May 28, 2016, 12:40:28 AM »
Test for randomness in such a small sample is not possible.  The randomness test are like they can tell a very high or low probability the sample is random, it can never say it is absolute sure random.

I'm not talking about small samples. I just can't post here thousands of outcomes.
In fact my point has nothing to do with statistics. It is based on pure logic. If something is random it means it is unknown. It does not follow rules. If it followed rules then those rules would limit its randomness. If it is unknown and does not follow rules then I can not evaluate it.
 

Jesper

  • Hero Member
  • ******
  • Posts: 1138
  • Thanked: 535 times
  • Gender: Male
Re: A Test for Randomness
« Reply #7 on: May 28, 2016, 12:59:21 AM »
No we can not know, as everything is possible. We use some rules which in most of the cases fit. As 400 spins from a casino whatever the numbers are they have the same probability as it were 400 zeros. Nobody would think 400 zeros is a random outcome, the wheel should be called have bias for much less. If it shows the same 400 numbers as last night, it should be a situation which should be unlikely as than a spacecraft were landing with aliens.

 

kav

  • www.Roulette30.com
  • Administrator
  • Hero Member
  • *****
  • Posts: 1531
  • Thanked: 581 times
  • Gender: Male
Re: A Test for Randomness
« Reply #8 on: May 28, 2016, 01:16:10 AM »
You make a good point and I may agree with you, but I want to prove a point.
And this point is that strictly speaking, when you start putting limits on randomness, when you start having "expectations", then this is the basis of (what people call) gambler's fallacy.

In your post, what you did is that you looked in past spins (400 zeros) to evaluate the next spins. In principle  this is "gambler's fallacy".
If one says that a wheel that consistently shows black numbers is not random, then one basically adopts a gambler's fallacy approach.
I'm not taking sides. I'm just pointing out a contradiction: If one can "test randomness" then he actually uses some form of "gambler's fallacy". One can not at the same time be against gambler's fallacy and "test randomness".

If you have expectations from the wheel (expectations of randomness, equilibrium, probability or whatever) then you can test and bet according to those expectations.
If you have no expectations from the wheel, then you can not bet according to them and can not test for randomness.
« Last Edit: May 28, 2016, 01:41:41 AM by kav »
 
The following users thanked this post: Reyth, Sheridan44

Jesper

  • Hero Member
  • ******
  • Posts: 1138
  • Thanked: 535 times
  • Gender: Male
Re: A Test for Randomness
« Reply #9 on: May 28, 2016, 02:47:16 AM »
Kav!  I did not mean 400 of reds twice in a row, it was any 400 numbers. Any 400 numbers has the same probability.

We see almost never even 50 equal sequence's.  That's why Dobbelsteen is right when he say we do not need to wait for ten reds, we can bet against the last ten as well.

I think it is still more to research in the random math, we probably not know it all.

 

Sheridan44

  • Mature Member
  • ***
  • Posts: 226
  • Thanked: 185 times
  • Gender: Male
Re: A Test for Randomness
« Reply #10 on: May 28, 2016, 03:36:50 AM »
Jesper, the way I see it is given enough time, in theory - one could almost bet against any sequence, it need not be "abnormal". Examples....RRRRRRR bet BBBBBBB...RBRBRBR bet BRBRBRB....or something where there seems to be little or no pattern like RBBRRBR bet the exact reverse - BRRBBRB etc... given time any pattern nearly becomes "unique". And the casinos know this - this is why they must have rather narrow betting ranges.

LOL - think of it. In our casino of the imagination, some bum could wander in from off the street - start with a five dollar bill - and play at this dream table (with a one cent minimum and an infinity maximum), and own the damn place within a couple of days.

Now saying all this is one thing, trying to develop a betting scheme to exploit it is quite another. Maybe Warren Buffett or Donald Trump could do it for awhile (especially Donald - he could set his own table limits...at his own casino that is)....LOL.
« Last Edit: May 28, 2016, 04:15:09 AM by Sheridan44 »
 

Reyth

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3014
  • Thanked: 798 times
Re: A Test for Randomness
« Reply #11 on: May 28, 2016, 06:59:11 AM »
LOL - think of it. In our casino of the imagination, some bum could wander in from off the street - start with a five dollar bill - and play at this dream table (with a one cent minimum and an infinity maximum), and own the damn place within a couple of days.

Ya, its much harder than it sounds.  These kinds of tables prove that variance is the largest enemy that we face and that it is in the casino's favor.
 

UK-21

  • New
  • **
  • Posts: 115
  • Thanked: 38 times
  • Gender: Male
Re: A Test for Randomness
« Reply #12 on: May 28, 2016, 07:47:37 AM »


I think the speech balloon should say:

"We know the wheel's perfect - get on to the DM and see if we can offer him a line of credit  . . . "

or

"Would you accept a suite for the night, on the house sir? It'll keep you from having to leave early to travel home."

Any others? We should have a competition?

 
The following users thanked this post: Sheridan44

Jesper

  • Hero Member
  • ******
  • Posts: 1138
  • Thanked: 535 times
  • Gender: Male
Re: A Test for Randomness
« Reply #13 on: May 28, 2016, 08:19:10 AM »
It is at least one on line casino having 1-200000 spread on an EC, and I know casinos claim no bet limits. They must know this will not help the players win over them.   The fair odds games has the same max, but 5 times higher min, which  is a smart way of hide it is fivefold cut.

They know it may be much more winning sessions if any can double up more times, but at the end the loss will be larger.  Casinos love Martingaler as they risk very little. When the Martingaler win on the last stage with about 48%
certainty he had already by him self paid in all the money except the first bet value.  We can double up only seventeen times, and risk near 4000 Euro for a cent.  The casino need just one cent to stand a Martingaler starting there and he can double up as long he wants, it is one cent for the casino.

The table max is not for stopping a Martingaler, it is to stop somebody to place a bet which the bank should have
problems to pay out (as 100000000 on black once). 

The variance together with the HA can kill any player. On a fair odds table has i theory that part which highest bank the best chance to win in the long run, usually is not the player that part.
It will allways be a game, not free money!

 
 
The following users thanked this post: Reyth

UK-21

  • New
  • **
  • Posts: 115
  • Thanked: 38 times
  • Gender: Male
Re: A Test for Randomness
« Reply #14 on: May 28, 2016, 08:22:49 AM »
New article :  A Test for Randomness [nofollow]

The creative system designer will see some uses for this, I'm sure.  ;)

I've read through your piece - interesting. I think it would be useful to add a snapshot of the sector of the wheel with the six numbers showing to add some context to the example; it's not immediately clear with simply the words and numbers (IMHO).

It's a long, long time since I did any heavy duty statistical sampling work (well over 20 years) but your piece has prompted me to do some further reading - I'm particularly interested in how this randomness testing concept can be overlaid with win/loss probabilities to arrive at a %age confidence level for randomness in results. I've had some eyebrow raising results when playing online recently, and last night the alarm bells started ringing when the frequency of a single number (over a 550-ish sample) broke into the four standard deviation sector of the bell curve. All of the others are not too far off of the EV.

 
The following users thanked this post: Bayes