Can the SB chart be distinguished from the price chart? - page 7

 
Maxim Dmitrievsky:

If no constant components can be distinguished, then of course the process is random, what is there to think about?

As for the price series, the trend is a constant component - well, it exists no matter how you argue, but "If no constant component can be singled out":

- there are trends, but only on history.

- we can't find a method to predict future trends.

and? are price series random? we can't single out a constant component - a trend? - Even seasonal fluctuations in price series are not predictable, the hell with currencies, even commodity markets.

imho, if market participants have information, it means that there is a trend, but if there is no information, it will be a random fluctuation, i.e. there are two mathematical models that replace each other in price series


This topic regularly appears on this forum, I usually look for Prival's posts, he even posted a patent study on "random process definition" somewhere, you should search forhttps://www by his posts.mql5.com/ru/search#!keyword=%D1%81%D0%BB%D1%83%D1%87%D0%B0%D0%B9%D0%BD%D1%8B%D0%B9%20%D0%BF%D1%80%D0%BE%D1%86%D0%B5%D1%81%D1%81&module=mql5_module_forum&author=Prival&page=1

Поиск - MQL5.community
Поиск - MQL5.community
  • www.mql5.com
Поиск выполняется с учетом морфологии и без учета регистра. Все буквы, независимо от того, как они введены, будут рассматриваться как строчные. По умолчанию наш поиск показывает страницы...
 

I used to draw charts with GPS. I used to flip a 'coin' - if it was 'heads' a tick up, if it was 'tails' a tick down, then I would form minute bars from such ticks and put them into MT. In purely visual form this turned out to be a load of crap - the price was flying like crazy, which obviously does not happen in reality. By experience I have found that a graph of such SB visually looks like a real price diagram, if 70% of ticks are related, i.e. if a current tick is up then the next one is down and vice versa. Such a graph becomes especially realistic, if the number of generated ticks is modulated according to changes of daily volatility.

In this case the distribution form becomes as it is drawn by danminin orVizard_ in the upper picture.

But it is of no use.
 
Igor Makanu:

as far as price series are concerned, the trend is a constant component - well, it is there no matter how much you argue, but "If no constant component can be distinguished":

- there are trends, but only on history.

- we can't find a method to predict future trends.

and? are price series random? we can't single out a constant component - a trend? - Even seasonal fluctuations in price series are not predictable, the hell with currencies, even commodity markets.

imho, if market participants have information, it means that there is a trend, but if there is no information, it will be a random fluctuation, i.e. there are two mathematical models that replace each other in price series


This topic regularly appears on this forum, I usually look for Prival's posts, he even posted a patent study on "random process definition" somewhere, you should search forhttps://www by his posts.mql5.com/ru/search#!keyword=%D1%81%D0%BB%D1%83%D1%87%D0%B0%D0%B9%D0%BD%D1%8B%D0%B9%20%D0%BF%D1%80%D0%BE%D1%86%D0%B5%D1%81%D1%81&module=mql5_module_forum&author=Prival&page=1

What the fuck are the layovers and the pile-ups. Quotes are random, it's random rambling. That's what this thread is about.

there is no such thing as being a little off, or sometimes off and sometimes not
 
Олег avtomat:

one in the woods and one out of the woods...

I highly recommend reading it:

Ventzel E.S., Ovcharov L.A.

Probability Theory and its Engineering Applications

Moscow: Nauka. 1988 (Physics and Mathematics Library for Engineer). - 480 с.

Exactly according to Wentzel, they taught statistics at our university, and I still tried to "fit" the course of prices into the Procrustean bed of the normal distribution. No way. We either lower the significance level to 0.5 or lower (but who needs such a low significance), or steadily reject the null hypothesis that the price movement is described by the normal or uniform distribution (I investigated only two distributions, but I think that checking all other distributions will lead to the same conclusion).

 
Maxim Dmitrievsky:

Quotes are random, it's a random stray.

of course not )
 
danminin:

It's not just that. It's the tails. But there's not that much difference.

is it?

it's in log scale. Let me explain, in the picture there is a two-sided exponent in red and a price distribution in blue, the two-sided exponent has heavier tails and high kurtosis, however, we can also see in this case that the tails of the price distribution are outside this distribution too, let alone the normal one.

 
TheXpert:
of course not )

How so? Non-periodic cycles do not count

 
Dmitry Fedoseev:

How do you determine that this is an entirely different process?

Dimitri, come on, don't "poke fun".

There is a section of statistics - hypothesis testing. We have a sample (I checked both price increments with each tick, and just price values with each tick). And then there's the standard test, whether this sample meets the normality criteria of the distribution. We accept the null hypothesis and the alternative hypothesis about the nature of the distribution, calculate the criterion, and reject one of the hypotheses. I mostly used the Shapiro-Wilk criterion as I mostly tested for normality. I also triedChi-Square test for uniformity and normality, and the null hypothesis (of uniform or normal distribution) was also rejected.

 

A successful trader is looking for investors, not padawans, so what's the point of reading the brainwashing excerpts of some Prival?


 
Maxim Dmitrievsky:

A successful trader seeks investors, not padawans, so what is the point of reading excerpts from the brain mush of some Prival?


Well, this success story is repeated by everyone, I can find another 5 people who used to be successful and then started to teach or look for investors.

he was interesting because he was mathematically "savvy", he used to teach at university, and in his posts there is a method (scientifically sound and probably defended by a dissertation?) that spells out how to determine a random process on five pages

I'll search in the evening, I don't have internet at the moment.