![MQL5 - Language of trade strategies built-in the MetaTrader 5 client terminal](https://c.mql5.com/i/registerlandings/logo-2.png)
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
It seems to be all according to science.
The range is from 0 (first difference series) to 1 (linear trend on large TF). The special place is occupied by a random Brownian one-dimensional motion (integrated SV with zero MO), for it PC=1/2, and a noisy sine, at this comrade, PC smoothly oscillates that should be, as on small TF the noise plays a large role, on large TF the trend is already visible, etc.
The PC for Y2 gets below zero.
Are you kidding me?
If you're serious, maybe consider, as an option, the statistical scatter of the studied value. Simply, on large TF, the number of samples in the series under study drops as 1/TF, hence the scatter growing as SQRT(TF), and given that the PC for the first difference always tends to zero as 1/SQRT(n), you can understand where the minus comes from in places.
Are you kidding me?
Well generally no.
If you're serious, maybe consider, as an option, the statistical scatter of the studied value. Simply, on large TF, the number of samples in the studied series drops as 1/TF, hence the scatter growing as SQRT(TF), and given that the PC for the first difference always tends to zero as 1/SQRT(n), you can understand where the minus comes from in places.
More on that point, please.
In the sense of the PC there should not be a single datum for which the condition R < S is satisfied.
Visually -- for Y2, the R/S plot should be greater than zero because there is noise, and the R/S plot should go up to 30, after 30 horizontally
Here's what might happen.
In the formulation that Prival implemented , PC is considered an integral exponent because it is defined through the tangent of the slope of the line drawn through the set of points. There are regions on this set with a negative slope, but in general (integrally), the slope is positive and there really can be no case where PC < 0.
The slope angle is calculated locally, between every two adjacent points and sometimes the spread is smaller on a larger TF and it happens... In this case "my" PC gives minus. In fact, there is nothing inappropriate about it, if we understand what is going on and, of course, everything depends on how we define the AP itself. It seemed to me more informative to output this indicator locally.
In general, this needs to be sorted out. By definition, XP shows the rate of increase in volatility of BP with an increase in TF. I built my algorithm based on this definition. But, one can see that it does not coincide with the original one or I am missing the point somewhere.
P.S. And then I have not got anything reasonable by formulas from the article (that Prival is glowing), I screwed up there (well, or in my head). Therefore, I would not appeal to expressions from there, as the truth.
I've also had negative values, I don't remember which ones, but I have. It jumps around a lot (which is why I didn't like it). I will try to compare two algorithms, your Neutron and mine.
TheXpert regarding N and n. If you insert N, X(N) will always equal zero. But I'll double-check, there's something wrong, this is where it becomes integral.
TheXpert regarding N and n. If you insert N, then X(N) always equals zero. But I'll double-check, something is wrong there, this is where it becomes integral.
Ha, that might be the mistake.
For a particular N there should be N - 1 values of X :
X[i] = Summ(i)(e[i] - M[N]) i = 2...N I hope this is clear
_______________________________
At least as it stands now, the expression certainly doesn't make sense -- calculating the cumulative deviation from MOG by N for n (i.e. all!) elements!
....
In general, this needs to be dealt with. By definition, the PC shows the rate at which the volatility of the BP increases with the TF. I built my algorithm exactly based on this definition. But, one can see that it does not coincide with the original one or I am missing the point somewhere.
P.S. And then I have not got anything reasonable by formulas from the article (that Prival is glowing), I screwed up there (well, or in my head). Therefore, I would not appeal to expressions from there, as the truth.
I too do not yet have a clear version of how to count it correctly. In different sources it is different. It is evident that these articles were written not by programmers. And take away from this "with the increase in TF", only confusing. It is the change in the water level of the river Nile, or the number of crocodiles. Once we calculate it properly, then we will think about what happens to it when the TF increases.
Here's what might be the case.
The slope angle is calculated locally, between every two adjacent points and sometimes it happens that the slope at a larger TF is less spread; it may happen so... In this case "my" PCB honestly bounces to minus. In fact, there is nothing inappropriate about it, if we understand what is going on and, of course, everything depends on how we define the AP itself. It seemed to me more informative to output this indicator locally.
Yeah, it's starting to make sense now.
In general, it is necessary to sort it out.
Uh-huh
By definition, the PC shows the rate of increase in BP volatility with increasing TF. I built my algorithm exactly based on this definition. But I see that it does not coincide with the original one or I am not getting it somewhere.
Maybe I should build it for a sinusoid without noise and compare it with the image in the article. So, let's ignore formulas from the article and take pictures as truth.
By the way, why don't you compare your values with the script?
I've had a blast today. The analog of Hurst's coefficient is possible to calculate quite locally!!!!!!!!!
This follows from Dubovikov's paper "Minimum coverage dimensionality and local analysis of fractal time series"
I've had a blast today. The analog of Hurst's coefficient is possible to calculate quite locally!!!!!!!!!
This follows from Dubovikov's paper "Minimum coverage dimensionality and local analysis of fractal time series"
Everything has already been stolen before us, hooray.