Fourier-based hypothesis - page 6
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
grasn: Спасибо! Поеду сегодня в отпуск - попробую повторить расчеты. А вдруг - Грааль? )))
Without the identification of the model, there is no way to get the grail. And with it, it won't be a grail by definition. But it could be a good system, theoretically.
And I need a library on linear algebra. By the way, you promised to help :o)
By the way you promised to send me the ToR.
I've already downloaded some books from the net to update the topic in my head.
(Linear algebra is a whole section of VM, I hope you don't think I will describe all VM in code :o)
By the way you promised to send me the ToR.
I've already downloaded some books from the net to update the topic in my head.
(Linear algebra is a whole section of VM, I hope you don't think I will describe all VM in code :o)
I don't need ToR, I even can't figure out how to write it. And I need an MQL implementation:
- matrix inversion
-matrix product
-transition of matrices
You don't have to write it all. There are ready-made dlls that are being used, I can't figure them out. Here, for example
https://forum.mql4.com/ru/4988/page4#96492 (Ilnur's very first post on this page). It's very confusing there. I found this clapck.dll library, but it has no specification, fuck, I searched everything, what to do with it - I don't understand. Went to LAPACK description - everything is cool, but how to use these driver routines (like ready-made solutions) in practice - I don't know.
How the matrix is applied I more or less understand, but how to make it work with mql is a mystery to me. It's easier to write a grail (this is kind of a joke).
MQL implementations are needed:
- matrix inversion
-production of matrices
-transpose matrices
This already sounds like a ToR. Is that all (what you need to be happy)?
I will not modify Lapak since it has the Object Oriented Approach in MQL-5,
and with 4 it's easier to re-write.
That already sounds like a ToR. Is that all (what you need to be happy)?
I won't remake Lapak, because it has the object-oriented approach in MQL-5,
but on the 4 is just a load of crap, it's easier to rewrite.
I think so. I don't think I missed anything.
I didn't dig deep into the maths, but let's assume it's true.
...
The 1st segment is one impact, the 2nd is completely different, and we're kind of adding them up here. :)
Where do we fold? Who and with what?
Do you even know mathematical symbols or are you a total lamer?
Show me at least one addition sign when calculating amplitudes or phases in the example I have given.
There is a hypothesis: If we take a segment of prices, suppose for the last 1000 bars, and approximate it by FFT, then if we correctly capture the basic harmonics by FFT, we can equally extrapolate prices not only into the future, but also into the past.
This can be done, for example, as follows: we can select such a set of FFT parameters (number of harmonics, approximation accuracy) so that it would give the minimum RMS at the interval preceding the selected one (for example, from 1200 to 1000 bars). In this case there is a probability that the selected coefficients will approximate not only the previous interval, but also the future one from 0 to 200 bar (of course, if the basic market rhythms do not significantly change).
Colleagues, can anyone help to test the hypothesis?
Hi! Digged a lot in similar directions, only did harmonic sampling on a test sample just before the future. The test sample didn't naturally fit
into the Fourier decomposition. I kept those harmonics, which in total gave the best correlation with the initial series. Sometimes it turned out to be a beautiful prediction, and sometimes it turned out to be a total crap.
The problem is that there is no criterion for assessing the quality of the forecast, when you cannot see the future in real time. Therefore the value of such a predictor is questionable.
Yes, for the forecast using this method it is important that the "instantaneous" Fourier spectrum is stable on the implied horizon, but it changes on every new bar. Increasing the decomposition window
gives false constancy effect to the spectrum which, by definition, is not "instantaneous" and reflects all the harmonics in this window, but none of the harmonics that
take place at the boundary of the past and the future, much less in the future itself. The Fourier decomposition is rather an approximation of the window, which knows nothing of what is going on outside it,
and in no way links the past and the future in a common model. If you finish drawing a woman at the end of the decomposition window, it will also decompose that window into harmonics, thereby correcting all the Ak,Bk - coefficients to account for the new finish drawing.
I hope I was of some help...
Yes, all this really does not need to be written. There are ready-made dlls used, I can not figure them out. For example, here
https://forum.mql4.com/ru/4988/page4#96492 (Ilnur's very first post on the page). It's very confusing there. I found this clapck.dll library, but no specification for it, fuck, I searched everything, what to do with it - I don't understand. Went to LAPACK description - everything is cool, but how to use these driver routines (like ready-made solutions) in practice - I don't know.
I more or less understand how the matrix is applied, but how to make it work with mql is a mystery to me. It's easier to write a grail (it's like a joke).
Here I gave an example of implementation of matrix reversal algorithm in MQL (taken from LAPACK library source codes).
Where do we fold? Who and with what?
Do you even know mathematical symbols or are you a total lamer?
Show me at least one addition sign when calculating amplitudes or phases in the example I gave you.
Good continuation of the discussion. Keep up the good work.
I found this clapck.dll library, but it has no specification, I searched everything, I don't know what to do with it.
I found in my "stash" a header file with matrix reference algorithm implementation based on external library clapack.dll.
I also added the library itself to the archive, just in case.