Dependency statistics in quotes (information theory, correlation and other feature selection methods) - page 67
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Okay, I'm going to bed, guys. I couldn't sleep for three hours - I suspect it's because of this thread. I'll crawl out here in the afternoon.
Let's get back to the topic at hand.
The article referred to by the topic starter has one significant flaw for me: there is no comparison of the calculation of the information dependence with the ACF. Let's do the math. I'll take the ACF calculation if they give me a .csv.
Whogetsch on the five I don't know. However, I can see that he came across a very interesting result when researching and didn't see it.
What didn't he see? Do you mean that the distribution of the number of ZZ tribes is like a graph of the distribution of cities according to their numbers?
In fact, the distribution for a ZZ is a property of the ZZ algorithm itself. Construct it on any non-stationary series and you will get the same distribution of knees. Therefore it is not necessary to draw any conclusions from this distribution
Let's get back to the topic at hand.
The article referred to by the topic starter has one significant flaw for me: there is no comparison of the calculation of the information dependence with the ACF. Let's do the math. I will undertake to calculate the ACF if I am given a .csv.
Thank you for reading my article more than once.
I will give you the data for calculation in this thread, and then let's see how it can be compared.
PS: In the meantime I will create a very interesting separate topic for coryphaei and amateurs ;)
getch 01.04.2010 11:03
Repeat:
EA counts the number of ZigZag knees (at least Pips) and writes to file:
EA text
The graph as a function of the number of elbows on their min. size (Pips):
Let's get back to the topic at hand.
The article referred to by the topic starter has one significant flaw for me: there is no comparison of the calculation of the information dependence with the ACF. Let's do the math. I am willing to calculate the ACF if I am given a .csv.
What didn't he see? Do you mean that the distribution of the number of knees of ZZ is similar to the graph of the distribution of cities according to their number?
In fact, the distribution for a RZ is a property of the RZ algorithm itself. Construct it on any non-stationary series and you will get the same distribution of knees. Therefore it is not necessary to draw any conclusions from this distribution.
If you follow the link to the "cognitivist" website and read it, you will see what the properties of the fractal series distribution are.
The matter is not in the algorithm, or rather not exactly in it. It is about the formulation of the problem and the goal. The aim is to gain profit. We shall gain profit until the price turns against our position. This is the entire algorithm. Now we should investigate how to maximize the profit. It can be maximized by investigating the structure of the engine, by decomposing into ranges and getting statistics of conversions to a higher (or lower) distribution.
If it is possible to get such distribution on "any non-stationary series", it means it is stable, so there is an order to profit from this stability. And it is necessary to draw conclusions.
HideYourRichess:
I do not know what charts were used but when I tested ZZ constructed by Pastukhov they did not show the same results. As a result, I concluded that the price character changes depending on the scale. To the opposite. On the other hand, if we operate with data not lower than some limit, we can disregard that "opposite part" and then yes, it is like fractality and so on.
Can you be more specific?
The algorithm is stated in this sentence
The Expert Advisor counts the number of ZigZag knees (not less than Pips) and saves them in the file
I'm sorry, I did not look at the Expert Advisor's code but it follows from this sentence that the number of passes for calculation of the number of bends should be equal to the number of pips on a one-minute timeframe from the maximum price range over the history.
If you follow the link to the site "cognitivist" and read it, it will become clear what are the properties of the distribution of fractal series.
It's not about the algorithm, or rather it's not really about the algorithm. It's about setting the objective and the goal. The goal is to make a profit. We gain profit until the price turns against our position. This is the entire algorithm. Now we should investigate how to maximize the profit. It can be maximized by investigating the structure of the engine, by decomposing into ranges and getting statistics of conversions to a higher (or lower) distribution.
So if such distribution can be obtained on "any non-stationary series", it means it is stable, and God forbid to get profit from this stability. And it is necessary to draw conclusions.
have you got it yet? It does not follow from that distribution that there is any possibility of profit dilution.
I don't know what kind of PP was used there, but when I tested Pastukhov's PP, the graphs did not look like that at all. As a result, I concluded that depending on the scale, the nature of the price changes. To the opposite. On the other hand, if we operate with data not lower than some limit, we can disregard that "opposite part" and then yes, fractality and so on.
What is Pastukhov's ZZ? Pastukhov studied kagi/renko in a classical construction. This rule (2H) doesn't apply exactly to a GZ. There is a dependence on the value of the knee at points
It doesn't follow from that distribution that there is an opportunity for profit dilution.
I'm "you" to me, too, if there are no objections.
Why not? Is there a justification?