Volumes, volatility and Hearst index - page 10
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
No one minds if you check Yurixx's conclusions. That is, either repeat the first principles calculation he did or get the result analytically. Actually, as discussed earlier, all that is missing is a formula linking the spread to the standard deviation.
research on the spread distribution https://www.mql5.com/go?link=http://www.mathnet.ru/php/getFT.phtml?jrnid=sm&paperid=3245&what=fullt&option_lang=rus There seems to be a formula 2.14 for first and second momentum, but something doesn't seem to add up :)
P.S. https://www. mql5.com/go?link=http://83.149.209.141/php/getFT.phtml?jrnid=sm&paperid=3415&what=fullt&option_lang=rus continuation
Take the textbook "Introduction to Probability Theory" by Kolmogorov. There you will find the formula for the mean run in random walk.
I think there, as well as in wikipedia, it is about the standard deviation. Can you give a quote here, with a link to the page?
Therefore this statement
Ask Yurixx'a to calculate Hurst for series N*N from 0 to 1000 .
By the way, the proportionality between mean run, mean run and spread means that the curves are parallel in log-log coordinates. That is, Yurixx 's graphs clearly show that there is no proportionality between the RMS run and the spread. Of course, if its calculation is correct. But between the average run (i.e. the Open - Close module) and the spread it is possible.
I don't know exactly how Yurixx calculated it, but the result:
Для небольших значений величины интервала N показатель существенно отличается от 0.5 и только с ростом N стремится к 0.5, повидимому асимптотически.
Exactly the same as I got three years ago for the main quotes. And I won't even check it. My result was a complete revision of the principles of the system. The only difference is that I came to the humble conclusion that TA does not work at all.
Vita, too many words and not enough specifics. You don't make any reference, and you don't draw your own conclusions. In addition, you constantly use all kinds of terms, as well as expressions (High-Low), (Open-Close), confuse them with each other and establish completely arbitrary ungrounded connections between them. And concerning the Hurst Ratio you are mistaken in general, about what it is and how to calculate it.
If you want to debate, then speak substantively: definition - assertion - proof - result. Or cite specific places from other authors. It would also be good to understand what I have written here. I doubt that you have understood it.
isceldating the spread distribution https://www.mql5.com/go?link=http://www.mathnet.ru/php/getFT.phtml?jrnid=sm&paperid=3245&what=fullt&option_lang=rus There seems to be a 2.14 formula for the first and second momentum, but something doesn't seem to add up :)
Z.I. http://83. 149.209.141/php/getFT.phtml?jrnid=sm&paperid=3415&what=fullt&option_lang=rus continuation
As far as I see, everywhere the first momentum of the spread is proportional to the root of T on the interval [0;T]. В
Also proportional to the root of T is the average run.
This allows us to assume that High - Low = k * |Open - Close|.
|Open - Close| is the average run.
Vita, too many words and not enough specifics. No reference in any meaningful way, no conclusions of your own. Besides, you constantly use different terms, as well as expressions (High-Low), (Open-Close), confuse them with each other and draw completely arbitrary and groundless connections between them. And concerning the Hurst Ratio you are mistaken in general, about what it is and how to calculate it.
If you want to debate, then speak substantively: definition - assertion - proof - result. Or cite specific places from other authors. It would also be good to understand what I have written here. I doubt that you have understood it.
Especially for you, Yurixx, I give an analysis that leads you to the result in table 2b, justified by theorems about SB :
Further to the text of my first post:
With random walks, the average run is proportional to the square root of the number of steps. So the result of the calculation a la Hurst, reduced to h = Log(High-Low)/Log(N) or something like that, after applying simple arithmetic, reveals the following:
1) High - Low = k * sqrt(N);
2) h = log (k * sqrt(N)) / log (N);
3) h = 1/2 + log(k) / log (N);
4) h -> 1/2 at k << N, which the table perfectly confirms.
As you see, I emphasize again, there is no Hurst here. There is the topiccaster formula and the mean run theorem for SB, which leads us directly to the result of Table 2b. The result of this table has no Hearst properties due to the incorrect nature of the original formula. For example, High - Low > N this formula does not digest, as it is adapted to obtain some result less than one only in the artificially constructed series of the same author.
My assessment of your results is more rigorous and in exact agreement with your data, without caveats like "should, but I don't know how to fit" and other of your lyrics about Hearst.
And more specifics about Hearst (see attached file). This is how I do my R/S analysis in it, which can count Hearst for any case, including the N * N series.
I've given the analytics to explain your h>1/2 for SB and the calculation of the Hearst figure, which, by the way, doesn't need to be tucked into an artificially invented series so that it doesn't screw up.
It's possible that you're confused by what I've written. Or you can't quite keep up. Then please forget my conclusion and the attached file for Hearst's calculation. Suppose, in that case, I was talking nonsense. You don't have to understand it.
Show you yourself that your formula calculates Hearst.
Can you show me Hearst's calculation for a series N * N according to your formula? Or does your formula calculate your Hurst only for your series? Can you give the analytical output of the simplified case for your series?
Maybe you can even give the analytical derivation of your Table 2b results explaining h>1/2 instead of writing such "specifics":
Theoretically, for the SB in question, the Hurst figure should have been 0.5. However, as we can see, this is not the case.
To me it is obvious that this You have is not observed. Everyone who knows how to calculate Hearst observes the consistency of numerical calculations with theory. And for SB Hurst erases to 1/2 not only from above.
research on the spread distribution https://www.mql5.com/go?link=http://www.mathnet.ru/php/getFT.phtml?jrnid=sm&paperid=3245&what=fullt&option_lang=rus There seems to be a formula 2.14 for first and second momentum, but something doesn't seem to add up :)
P.S. https://www. mql5.com/go?link=http://83.149.209.141/php/getFT.phtml?jrnid=sm&paperid=3415&what=fullt&option_lang=rus continuation
Well yes, and for the distribution formula is 2.20, which refers to 2.13 :).
At least I was sure that people were studying the swing distribution and it's not that simple, here's a confirmation, thank you.
Yuri, here are the formulas :)
Yurixx, M is the average increment modulus over K intervals. It should increase in proportion to the root of N ?
So for example M(16)=M(4)*sqrt(4)
"the average distance from the starting point increases as the square root of time" (c) Einstein)))
Vita, look, you can't treat reality like that. You ignore that the root of T is only a funcion argument, to you Hurst, "reduced to h = Log(High-Low)/Log(N)". is "sort of."
More accurately, you can only do this if you are interested in something other than the truth in this discussion.
I guess I won't try to convince you anymore.
Vita, look, you can't treat reality like that. You ignore that the root of T is only a funucion argument, to you Hearst, "reduced to h = Log(High-Low)/Log(N)". is "sort of."
You can be more precise only if you are not interested in the truth in this discussion, but in something else.
I won't try to convince you anymore.
First of all not only the argument of the function, but also the multiplier of this function. For the numerical experiment that is carried out here in table 2b, the result of this function is constant, but we have already buried too deeply in search of the truth. Yes, and can you yourself directly say that High - Low = k * sqrt(N) is wrong?
It's much simpler than that. Calculate using the topcaster Hearst's formula for N*N*N. Or estimate the result relative to 1. What is the truth?
Maybe the artificially invented series under the formula h = Log(High-Low)/Log(N) is relevant to the market? Is the truth here?
The topikcaster invented the series, came up with the formula and declared it Hurst. Let him prove that it is Hurst if it is true. Kicking those who are gone and those who are far away is much easier.
...
The topikcaster came up with a series, came up with a formula and declared it to be Hurst. Let him prove it to Hearst if it is true. Kicking those who are gone and those who are far away is much easier.