Our members are dedicated to PASSION and PURPOSE without drama!

The Truth @Turner - Regression towards the mean

Started by Sputnik, June 21, 2014, 07:54:18 PM

Previous topic - Next topic

0 Members and 3 Guests are viewing this topic.

Sputnik

 
I crack the random flow.
I come up with a solution where the random flow has to show you the truth.
No bet selections or triggers.

With this solution you can observe how regression towards the mean unfold with out any fuzzy explanation.

Read this twice:

Regression toward the mean simply says that, following an extreme random event, the next random event is likely to be less extreme. In no sense does the future event "compensate for" or "even out" the previous event, though this is assumed in the gambler's fallacy (and variant law of averages). Similarly, the law of large numbers states that in the long term, the average will tend towards the expected value, but makes no statement about individual trials. For example, following a run of 10 heads on a flip of a fair coin (a rare, extreme event), regression to the mean states that the next run of heads will likely be less than 10, while the law of large numbers states that in the long term, this event will likely average out, and the average fraction of heads will tend to 1/2. By contrast, the gambler's fallacy incorrectly assumes that the coin is now "due" for a run of tails, to balance out.

What does this means:
This means i can pick any random combination of 10 events and compare them with the next 10 random events.
The expectation should be that they will not be the same combination and there will be present change.
So i can use opposite and same to see if this is the truth.

2
2
1
2
2
1
2
2
2
1

- - -

2 2 S
1 2 O
1 1 S
2 2 S
2 2 S
2 1 O
2 2 S
1 2 O
1 2 O
2 1 O

- - -

1 2 S
2 1 S
2 1 S
2 2 S
2 2 S
1 2 O
2 2 S
1 1 S
1 1 S
2 2 S

- - -

2 1 O
1 2 O
1 2 O
1 2 O
2 2 S
1 1 S
1 2 O
1 1 S
1 1 S
1 2 O

- - -

2 2 S
1 1 S
2 1 O
2 1 O 
1 2 O
2 1 O
2 1 O
1 1 S
1 1 S
1 1 S

- - -

2 2 S
2 1 O
2 2 S
2 2 S
1 1 S
1 2 O
1 2 O
2 1 O
2 1 O
2 1 O

- - -

2 2 S
1 2 O
2 2 S
2 2 S
1 1 S
2 1 O
1 1 S
1 2 O
1 2 O
1 2 O

- - -

2 2 S
1 1 S
2 2 S
2 2 S
2 1 O
2 2 S
1 1 S
2 1 O
1 1 S
2 1 O

- - -

2 2 S
1 1 S
2 2 S
2 2 S
2 2 S
2 2 S
1 1 S
2 2 S
1 1 S
2 2 S

- - -

What is the solution:

Lets assume you would use Ecart play and see windows of 3.0 STD as rare and extreme event.
Then for example 14 events vs 2 events is 3.0 STD.
That is a total of 16 events.

Then you can for example pick windows of 8 random trails following by 8 random trails and see regression towards the mean unfold.
Then when you get 2 events vs 3 events then the 3 events make the STD grow stronger and the 2 events is part of the regression toward the mean.
Now you can see what happens each time a window not grow stronger and hit 3.0 STD.
You do nothing then just observe the regression towards the mean.

2
2
2
2
1
1
1
1

- - -

2 S
2 S
1 O
1 O
1 S
1 S L
1 S L
1 S L 3.0 STD

- - -

2 S
2 S
1 S
2 O
1 S
2 O
2 O W 2.5 STD
1 S

- - -

2 S
1 O
1 S
1 O
1 S
1 O W 2.0 STD
1 O
1 S

- - -

1 O
1 S
1 S
2 O
2 O
1 S W 1.80 STD
2 O
1 S

- - -

2 O
1 S
2 O
2 S
2 S
2 O W 2.5 STD
2 S
1 S

- - -

2 S
1 S
1 O
1 O
2 S
1 O W 1.5 STD
1 O
2 O

- - -

2 S
2 O
1 S
1 S
1 O
1 S  L
1 S  L
1 O W 2.5 STD

- - -

2 S
1 O
2 O
2 O
1 S
1 S W 2.5 STD
2 O
2 O

- - -

2 S
1 S
2 S
2 S
2 O
1 S
1 O
1 O W 2.5 STD

- - -

3.0 STD No regression
2.5 STD Small regression
2.0 STD Small regression
1.8 STD Medium regression
2.5 STD Small regression
1.5 STD Large regression
2.5 STD Small regression
2.5 STD Small regression
2.5 STD Small regression

Development and experimenting:

8 random trails following by 8 random trails is only one example and can be any number.
You can pick what ever you think is extreme and rare event.

Comparing with this method then following is true.
Indication for regression event is the underrepresented events.
Indication for the STD to grow is the overrepresented events.

So if you for example see one red and two black, then red will part of  future regression as the underrepresented event and they can come in any combination
So if you for example see two red and three black, then red will be part of future regression as the underrepresented event and they can come in any combination.

Window and probability:

You can pick smaller windows around 2.5 STD or larger window up to 6.0 STD.
10 reds has the same probability as 10 in any combination, that is why you can pick any events with any combination with any length using the random flow comparing random against random.

The March
You have to develop a march if you are going to attack several times for regression.



Turner

Sputnik....


Thanks.


I have read it several times....and will read it again tomorrow. I have to admit that I am not fully getting it all.


RouletteFan

Really

great work Sputnik

Congratulation

thank you for sharing very interesting info

Best from Fance

Sputnik

Quote from: Turner on June 21, 2014, 09:37:04 PM
Sputnik....


Thanks.


I have read it several times....and will read it again tomorrow. I have to admit that I am not fully getting it all.

Lets say you have 10 red outcomes, then regression towards the mean says that the next 10 outcomes will be less extreme.
But it does not say it will even out with blacks 100% ...
So if i have 10 reds and no regression, then i would have 20 reds and that my friend is extreme and rare.


10 reds is rare and extreme, but 10 reds has the same probability as any random sequence with 10 outcomes.
This is why i can pick any sequence of 10 random outcomes and have the expectation that the next 10 random outcomes will not be the same 10 random outcomes.
And if i would see 10 same result, well then i would not placed any bets, as you only place bets when there is regression present or indication for present change.

That way i can see how regression towards the mean behave in the real world with real results with quick samples.
For example i don't need to wait for 3.0 STD window.

Now the beauty of this is that i can pick any window i think is extreme or rare.
Here is one example where i play for regression towards the mean, where the window should hit 2.30 STD or below and not higher ... that is 11/1 or 12/0 and i lose.
I can only play when i see two events growing and one event being regression component, underrepresented.

This is how the LW-Registry look like:
LWLWWLWWWW LLL WWWLWWLLWWLWWLWLWWWWLLWWWWWWLWLWWLWW LLL LWWLW

If you want to name this method for something else then regression towards the mean, then you can call it variance tracking.


2
2
1
1
1
1

-

1 2 O
1 2 O
1 1 S
2 1 O L
1 1 S W
2 1 S

-

1 1 S
2 1 O
2 1 O
1 2 O L
1 1 S W
2 2 S

-

2 1 O
2 2 S
1 2 O
1 1 S W
2 1 O
2 2 S

-

2 2 S
1 2 O
2 1 O
2 1 O L
2 2 S W
1 2 O

-

1 2 O
1 1 S
2 2 S
1 2 O W
1 2 O
1 1 S

-

1 1 S
1 1 S
1 2 O
2 1 O W
2 1 O
2 1 O

-

1 1 S
1 1 S
2 1 O
1 2 O W
2 2 S
2 2 S

-

2 1 O
1 1 S
2 2 S
1 1 S L
2 2 S L
2 2 S L

-

2 2 S
1 1 S
2 2 S
1 1 S
2 2 S
1 2 O

-

2 2 S
2 1 O
2 2 S
2 1 O W
1 2 O
2 1 O

-

2 2 S
1 2 O
2 2 S
1 2 O W
2 1 O
1 2 O

-

1 2 O
1 1 S
2 2 S
2 1 O W
2 2 S
2 1 O

-

1 1 S
2 1 O
1 2 O
1 2 O L
2 2 S W
1 2 O

-

1 1 S
1 2 O
1 1 S
2 1 O W
1 2 O
2 1 O

-

1 1 S
1 1 S
1 1 S
2 2 S
1 1 S
1 2 O

-

2 1 O
1 1 S
1 1 S
2 2 S L
1 1 S L
2 1 O W

-

2 2 S
2 1 O
1 1 S
1 2 O W
2 1 O
2 2 S

-

2 2 S
2 2 S
1 1 S
1 1 S
1 2 O
1 2 O

-

2 2 S
2 2 S
1 1 S
1 1 S
2 1 O
2 1 O

-

2 2 S
1 2 O
2 1 O
2 1 O L
2 2 S W
2 2 S

-

2 2 S
2 1 O
1 2 O
2 2 S W
2 2 S
2 2 S

-

2 2 S
2 2 S
2 1 O
2 2 S L
1 2 O W
1 2 O

-

1 2 O
1 2 O
2 2 S
1 2 O L
1 1 S W
2 1 O

-

1 1 S
1 1 S
2 2 S
2 1 O
1 1 S
1 2 O

-

1 1 S
2 1 O
2 2 S
1 2 O W
1 1 S
1 1 S

-

1 1 S
1 2 O
1 2 O
1 1 S W
2 1 O
2 1 O

-

1 1 S
2 1 O
1 1 S
2 1 O W
1 2 O
1 2 O

-

2 1 O
2 2 S
1 1 S
2 2 S L
1 1 S L
2 1 O W

-

2 2 S
2 2 S
1 1 S
2 2 S
2 1 O
1 2 O

-

1 2 O
1 2 O
2 1 O
2 2 S
1 2 O
2 1 O

-

1 1 S
2 1 O
2 2 S
1 2 O W
2 1 O
2 2 S

-

2 1 O
2 2 S
1 2 O
1 1 S W
2 2 S
1 2 O

-

2 2 S
1 2 O
1 1 S
2 1 O W
2 2 S
1 1 S

-

1 2 O
2 1 O
2 1 O
2 2 S
1 2 O
1 1 S

-

2 1 O
2 2 S
1 2 O
2 2 S W
1 1 S
1 1 S

-

2 2 S
2 2 S
2 1 O
1 2 O W
1 1 S
1 1 S

-

1 2 O
2 2 S
1 2 O
2 1 O L
1 1 S W
2 1 O

-

2 1 O
2 2 S
2 1 O
1 2 O L
1 1 S W
1 2 O

-

2 2 S
1 2 O
2 2 S
2 1 O W
1 1 S
1 1 S

-

2 2 S
2 1 O
2 2 S
2 2 S L
2 1 O W
2 1 O

-

1 2 O
2 2 S
1 2 O
2 2 S W
2 2 S
1 2 O

-

2 1 O
2 2 S
1 1 S
2 2 S L
2 2 S L
1 1 S L

-

2 2 S
2 2 S
1 1 S
1 2 O
1 2 O
2 1 O

-

2 2 S
1 2 O
1 1 S
1 1 S L
2 1 O W
1 2 O

-

2 2 S
2 1 O
2 1 O
1 1 S W
1 2 O
2 1 O

-

2 2 S
2 2 S
1 2 O
1 1 S L
2 1 O W
1 2 O

-


Sputnik

I am into this, math and probability and valid reason why to place bets.
You might call it mechanical, i call it tendency play.
We can not outguess 50/50 - that is why mechanical play is as good as any other existing method and tendency play fall between both.

Turner

Now I get it. Bravo.
Its beautiful. I see GrandMaster combinations in Chess    that are beautiful...almost art. This is is same.
Now I need to study other lengths and SDs
Well done. Well shared

Sputnik

Quote from: Turner on June 22, 2014, 06:43:44 PM
Now I get it. Bravo.
Its beautiful. I see GrandMaster combinations in Chess    that are beautiful...almost art. This is is same.
Now I need to study other lengths and SDs
Well done. Well shared

Exactly now we can see the regression and how it behave with any length or STD.
For example:

2
1
2
2
1
2
1
1

1

- - -

2 2 S
1 1 S
2 2 S
2 2 S
2 1 O regression
2 2 S
1 1 S
2 1 O regression
1 1 S
2 1 O regression

- - -

2 2 S
1 1 S
2 2 S
2 2 S
2 2 S
2 2 S
1 1 S
2 2 S
1 1 S
2 2 S no regression

The question is how to capture regression.
First observations show that is very common that you get at least two event as part of regression and more events.

When something is due we would never bet, so we can skip that part chasing for events.
But what is a indication of present change, one event show us present change, but would we attack after that?
For how many attempts should we attack, twice or three times in a row.

The chart of overrepresented and underrepresented events can give us a clue when to bet and when not to bet.
We have to observe how much stronger the STD grow and how it gets weaker.


Mike

Hello Sputnik,


It seems you are betting that the next 10 spins won't match the previous 10 spins, is that right?


1 R
2 R
3 B
4 R
5 B
6 B
7 B
8 R
9 R
10 B


If the next spin, no. 11 is R then it matches spin 1 which is R, so you put a S label on it, but if spin 11 was B, you would put a O label there. But I don't understand how you are choosing your times to bet. Is it when the outcomes (S or O) are starting to balance out?


1 2 O[/size]1 2 O1 1 S2 1 O L    Why did you bet here?1 1 S W2 1 S - 1 1 S2 1 O2 1 O1 2 O L1 1 S W2 2 S

Sputnik

Quote from: Mike on June 23, 2014, 08:43:46 AM
Hello Sputnik,


It seems you are betting that the next 10 spins won't match the previous 10 spins, is that right?


1 R
2 R
3 B
4 R
5 B
6 B
7 B
8 R
9 R
10 B


If the next spin, no. 11 is R then it matches spin 1 which is R, so you put a S label on it, but if spin 11 was B, you would put a O label there. But I don't understand how you are choosing your times to bet. Is it when the outcomes (S or O) are starting to balance out?


1 2 O1 2 O1 1 S2 1 O L    Why did you bet here?1 1 S W2 1 S - 1 1 S2 1 O2 1 O1 2 O L1 1 S W2 2 S

This is observation where the random flow can not fool you and has to show you the truth about regression towards the mean.
Tricky part is to capture it.

But lets say you have two same and one opposite in the beginning - then the two same make the STD to grow stronger and the opposite make the STD getting weaker and they can come in any combination.
That would be overrepresented events and underrepresented events.

The question is would you attack once, twice or three times ? and what happens next after that ...
I think you need to see what is common.
This is not a complete playing model, is more experimenting and observation about regression towards the mean.

RouletteFan

Hi Sputnik

can you explain how your graph work with a small exemple

how to read the graph to see what level standart deviation we reach
of course this sndart deviation go up or go dow depend on the next spins

thank

RouletteFan

One more thing

i have studied myself extensively  EC chance
i have never see a LW registry as you show

because even with the LW registry i have always encounter
long series of win and long serie of loss exactly as an Ec distribution
and if i take 10000 decision as an LW registry i will have 5000 L 5000W almost

all this follow the low of binomial probability as  you need t have 2 serie of 1 for on serie of 2 etc etc you have studie this with marigny

even with marigny as i have also work on it and play it live
you will have a registry of long decision loss long desision win  low serie etc
witch also follw the law of binomial probability

like LLWLLLWLWWWLWLLLLLWWLWWWWWLLLLLLLLLWWL

this kind of thing you cannot win

and remember if you have read the book

marigny say at one moment
when you have win a lot (mean you hve LWLWWLLWWWLWLWLWWWW
please decrease the value of your bet ??? ??

question

in extentsive test you have done

can you confirm you have this LW registry ???

LLWLWWLLWWWLWWLWLWLLWWWW etc where you have not serie of 4,5,6,7,8and more series of L
so you got a kind of winnig system

thank again for this great post

sqzbox

Mike - the answer lies in this statement from Sputnik (referring to the example he posted above).

QuoteI can only play when i see two events growing and one event being regression component, underrepresented.

Take for example -
1 1 S
2 1 O
2 1 O
1 2 O L
1 1 S W
2 2 S

The series is S-O-O so now we see that the O-O represents "two events growing" and the first S is the "one event being regression". So the bet would be S, which loses in this case. But we bet S again and win and this is the end for that group of 6. Then we copy the outcomes we had here (1-2-2-1-1-2) for the next 6 and watch again.

2 1 O
2 2 S
1 2 O
1 1 S W
2 1 O
2 2 S

O-S-O shows 2 O's and one S and so the bet again is S and this time wins - end for this group of 6.

And so on. Sputnik will correct me I am sure if I have this wrong.

I like this very much for several reasons - it does not try to subvert the math but rather flows with it; an attack is limited to a maximum of 3 losses; and a long unending series is always avoided.

This is what I would call both elegant and efficient.  Well done Sputnik.


Turner

Sputnik.....in the real world...as you are fully aware....we encounter zero
What is your view on receiving a zero during qualifying phase
1 1 S
1 2 O
2 0
1
1
2

or during prediction phase
1 1 S
1 2 O
2 1 O
1 2 L
1 0
2


And.....the continuation to compare has a zero in the 6

Sputnik


I ignore zero as it does not exist and continue the play.

Cheers

RouletteFan

hello

here is y LW registry  using this way of play describe by sputnik

i do it with 6 spin and seach the regerssion in the next 6 spins

LLWWLWLWWLLLLWLWLLLLWWLWLWW

sputnik do you have other LW registry to show us ?

is there anybody in this forum have made some test ?