Hearst index - page 7
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
looked it up. Again 25. there's a correlogram, it's a function. A function turns into a number only when a certain value of the argument is given.
"In time series analysis, a correlogram, also known as an autocorrelation plot, is a plot of the autocorrelations of a sample, from h (time lag). "
this is what it looks like 'Autocorrelation function' it is a graph !!!
Now what does the graph (function) get compared to a number ? so is it ?
Or maybe you just have to compare not a function but a number to a number.
Hearst index is a number and should be compared to a number !!!
Z.I. The correlogram and ACF are essentially a set of autocorrelation coefficients. It uses a single number "autocorrelation coefficient (one)". So I wanted to find out what it is, what do you think it is, at what value of the argument, the autocorrelation function becomes an autocorrelation coefficient. Some fix the ACF at 0.707, some through the integral - this is important for another problem. Determining the time interval during which a process is correlated with itself. (For traders, this is the time during which the observed process retains its motion characteristics).
The Hurst index (HH) is a number that characterizes a given BP. Now, let's take a quotient, for example M1, find the Hearst Ratio for it (as long as everything is correct and there is no logical error). Let's carry out the same procedure for M2, M3...Mtf and obtain a graph - dependence of PC on TF. We compare it, if necessary, with my correlogram (also a graph from TF).
All this is not necessary? Then we find the autocorrelation coefficient in the series of the first difference, for example M10 and compare it to the PC for the same M10.
Serguei, where are the inconsistencies? Everything is compared without contradiction - number to number, function to function!
The Hurst Score (HH) is a number that describes a given BP. Now, let's take a quotient, for example M1, find HF for it (as long as everything is correct and there is no logical error). Let's carry out the same procedure for M2, M3...Mtf and obtain a graph - dependence of PC on TF. We compare it, if necessary, with my correlogram (also a graph from TF).
All this is not necessary? Then we find the autocorrelation coefficient in the series of the first difference, for example M10 and compare it to the PC for the same M10.
Serguei, where are the inconsistencies? Everything is compared without contradictions - number to number, function to function!
1. making up your own function and calling it by the name of another well-known function is a bit incorrect. (Mathcadet has a built-in ACF function lcorr() - it's simpler and more convenient).
2. "...find the autocorrelation coefficient in the first difference series..." - How ? what is it ? the formula ? (Autocorrelation means comparing a series to itself, if without shifting, then correlation = 1 by definition, when shifting the coefficient can vary from -1 to 1). Unit all the time compare with PC ?
Sergei, maybe Skype is better, faster voice to explain everything + progami on Matkadam explain to each other what we are talking about. We will erase the keyboard here. Most likely just a confusion of terms. That's why we do not understand each other.
Sergey, maybe Skype is better, it's quicker to explain everything with a voice + we can use a matcad program to explain to each other what we're talking about. We'll erase the keyboard here. There is most likely just a confusion of terms. That is why we do not understand each other.
And what do viewers do then. I don't think so. It's better to continue in the same direction in the same place. I mean on the form.
Although you could become a listener. But they can't.
What's the audience to do then. No way. It's better to continue in the same direction in the same place. That is, on the forma.
Although you could become a listener. But they won't be able to.
OK, I'll post the results in the form of formulas and graphs. I understand the aim. Hearst and the correlation coefficient - these are fundamentally different things or concepts of the same order (only varying in different ranges). I just don't understand how to calculate the "autocorrelation coefficient". I can do it, but I can't; I can do the coefficient of correlation but I can't do it because I don't understand what it is.
2. "...find the autocorrelation coefficient in the series of the first difference..." - How ? what is it ? the formula ? (Autocorrelation means comparing the series to itself, if without shifting then correlation = 1 by definition, the coefficient can change from -1 to 1 when shifting). Unit all the time to compare with PC ?
We don't consider the unit - trivial case. The shift in the series of the first difference is always 1 and only 1! - We consider the correlation only between neighboring samples in the series of the first difference in a REAL TF. To obtain the correlogram, we vary ONLY the TF for the initial series.
This is a correct definition, there should be no misunderstanding.
No. It is better to continue in the same way in the same place.
I agree. It's better that way.
To get a correlogram, we are varying ONLY the TF for the original series.
perhaps, Prival, you are right. This is not a correlogram, the correlation coefficient between neighbouring samples in the series of first difference found for different TFs.
perhaps, Prival, you are right. It's not a correlogram, it's a correlation coefficient between neighbouring samples in a series of first differences found for different TFs.
And it confuses me too. If two arrays are compared, let's say one is M1 and the other is M5, of course. But the arrays should be of equal length. Suppose there are 20 values. It turns out that we are comparing the behavior in different time horizons. Minutes is 20 minutes and 5 minutes is 1h 40 minutes. That doesn't sound right either.
We assume that the series is stationary in the first approximation and that there is no appreciable difference in the estimates obtained from the BP section on which this estimate is made.
We assume that the series is stationary in the first approximation and that there is no noticeable difference in the estimates obtained from the BP section on which this estimate is made.
Is there a calculation of the Hurst index in Matcad (we need formulas in discrete form)?
So far I have only found this
File with approaches to time series analysis attached. These formulas taken from there.
There is no such function in Matkad.
What you quoted in your post seems to be true, except for the following (correctly so):
1. Stable trends or predictable behaviour of BP: Hu<1/2 or Hu>1/2 (antipersistence and persistence respectively).
2. Lack of stability or unpredictability of BP behaviour: Hu=1/2 (integrated CB with zero MO in the first-difference series).