Drazen i will agree with you that we disagree.
Clustering and Markow chains is complex things.

Cheers

Ok. We can agree that we disagree

Singles and series have same correlation as R vs B

And that hovering state is something inbetween like artificial pattern and can't represent right value as those 2 can by itself.

Quote from: Drazen on March 03, 2013, 08:48:05 PM
Ok. We can agree that we disagree

Singles and series have same correlation as R vs B

And that hovering state is something inbetween like artificial pattern and can't represent original value as those 2 can by itself.

True that singles and sereis is 1 in 2 same as clustering that is 1 in 3 with out any other possibility's.
Same as you apply the principal 1/3 - then you can not change the fact you only have 8 existing clustering patterns.
I also have the STD and ECART for does ...

Pretty simple when you know fact from fiction.

Quote from: Sputnik on March 03, 2013, 07:14:56 PM

So clustering series have a bias ?

I read about some math experts who explore the law of series and come to the conclusion that clustering series where slight more common.

So you have three possibility's, only three existing states, 1 in 3 with no other possibility's.
And you say that FTL has a slight higher strike ratio ...

Then Spike and others would be correct when they state that you be better of following the wheel.

Now why do you say it is a flaw when you have 1 in 3 - that is not a flaw - that is what it is.
But that the distribution clustering into 1 in 3 show a bias towards FTL is just that, slight more common and also not be a flaw.

So the conclusion on developing a march is to aim for FLT and hovering state or FTL and series of singles and not hovering state and series of singles.

That is the best way to explore it. As per thousands of sessions i tested before pen and paper.
"...aim for FLT and hovering state or FTL and series of singles and not hovering state and series of singles."
Basics is that one should aim for series always.


The "bias" comes from the way we identify the states.
Series of singles
11 2 1 22 ends in a series
Hovering
(and here is the trick)
11 2 11 -- one hovering state also ends in a series
11 2 11 2 11 -- two hovering states, ends in a series ...


We often name hovering state as "Isolated Single"
It should be also "Isolated Series"


2 11 2 11 2 1 2- so here we would have two isolated series but only one isolated single followed by one series of singles (that did not end yet).
*****************************************************************************
Even if we change the way we identify the hovering state to be = Isolated Singles + Isolated Series
we can not solve the discrepancy towards series of singles.

One other common flaw when someone starts to count states in a binomial distribution (markow chains) is in small detail: - is that we count ONE series of singles despite its length but do not do so regarding series of series. at the 3rd series in a row we count another series of series state. Causing the count to be much bigger.


The strategy would remain the same and it does not matter, as there is clear bias towards series of series??


Does the distribution of this states have the same frequency as 12 numbers???


Thanks for this discussion even if it went a bit off-topc


Cheers

Thanks MG you are so good with words - i like and respect you for that and your knowledge 

Cheers

Have anybody test this.  Jump in at a point then a lot of numbers have been choppy an EC without any series over 4, for a long time.
Test is then to see if we can get  long series of any of the EC. Say we go for 12 in a row. Starting a reversed martingale give us over 1000 trials before we lose the 1000 units. Any early success will gain good.

Playing reverse martingale, start with a streak of 8, stopped at a net of 125 units. Here I did not wait for an entry as the streak started on the third spin. If we remember we lose only one chip a trial, it should not be scary to wait for 7 or more.

Quote from: Ralph on March 05, 2013, 09:40:18 AM
Have anybody test this.  Jump in at a point then a lot of numbers have been choppy an EC without any series over 4, for a long time.
Test is then to see if we can get  long series of any of the EC. Say we go for 12 in a row. Starting a reversed martingale give us over 1000 trials before we lose the 1000 units. Any early success will gain good.

Yes some of us did.

You can do it two ways. To track any length serie as event opposing to it all chops, or track length serie by absence in number of spins.

I hope you know that outside EC-s have highest STD out of all bets  so cornering one such event and showing in some reasonable time after some point as you would like to, will require realy high (or low on other side, as you wish) probability to be sure... But as you are increasing probability for something, opposite to it you are face with decreasing number of chances...

Unfortunately all in all, doing this, won't be easy and smooth as it maybe can sound as an idea..

You will just have to find compromise with number of chances you can create, and probability you want to obtain in any case you decide for anything.

But you have bots.. you are programmer, if anyone can actualy explore this, then it should be you guys, right? 

Huh, making serie of 12 to show will be nice and realy interesting challenge dough.. 

Good hunt

Drazen