Spectrum derivative (or spectrum acceleration)

 

Actually the title of the topic speaks for itself.

Started here https://www.mql5.com/ru/forum/137042, interested in who has researched and how deep this issue is, also interested in phase observations in this case.

 
Not so much for me... ) Could you first clarify the meaning of these spectrum manupulations. Do you mean the power spectrum of the signal? the result of a Fourier series expansion? or do you mean spectrum and phase in a different sense?
 

Actually, at first I will try to use simple masks when constructing the spectrum and then change MA to anything I like.

I'd actually rather take a larger sample - 1024 bars at least, use it to build the entire MA spectrum from 1 to 1024, then build all possible combinations, then build a macdi from a macdi, also all possible combinations, etc. I wonder how much deeper expansion makes sense and which combinations are more accurate.

 
avatara:
(It's not like that for me. ) Could you first clarify the meaning of these spectrum manupulations. Do you mean signal power spectrum? Or do you mean spectrum and phase have another meaning?

I'll attach a file with the example of decomposition, but all in a row, the phase, yes, I am interested. I don't know how everything in the process of calculations of the order changes and whether it is correct and what to pay attention to - what is valuable, and here with this decomposition everything becomes visible, all frequencies before eyes, but it is not enough.
Files:
jlxprlihgml_1.zip  1022 kb
 
you seem to have a different meaning in the term spectrum... Good luck and harmony (without any hesitation) in your endeavours. )
 
avatara:
you seem to put a different meaning in the term spectrum... good luck and harmony in your endeavours. )


Perhaps you are not lacking either) the spectrum comes in colours - colour gradation from .... and up to ...., there is a signal spectrum (I'm not an expert here, though I have an opinion), there is a spectrum of frequencies.

I meant the spectrum of all macdies, the spectrum of macdies can be 2 macdies, 100 even ......, the acceleration of the spectrum of 2 macdies is a derivative of the series of differences between them (roughly speaking), then what do you think the acceleration of the spectrum of 10 - and macdies is?

 
trollolo:


Perhaps you don't have enough) the spectrum of colours comes in colour gradations from .... and up to ...., there is a signal spectrum (I am not an expert here, although I have an opinion), there is a spectrum of frequencies.

It is all the same, as colours and signals have frequencies, hence they can be decomposed into a spectrum of amplitudes.

trollolo:

I meant the spectrum of all macdies, the spectrum of macdies can consist of 2 macdies, of 100 even ......, the spectrum acceleration of 2 macdies is a derivative of the series of differences between them (roughly speaking), then what do you think the spectrum acceleration of 10 - and macdies is?

That's where you should have started, to avoid confusion. That is, you have suddenly thought of attaching the term "spectrum" to some manipulation of technical analysis indicators, calling it a method described by Grandpa Krylov in his fable "Quartet".
 
Reshetov:
That's where you should have started, so as not to confuse things. That is, you suddenly decided to add the term "spectrum" to some manipulation with technical analysis indicators, calling it a method described by Grandpa Krylov in his fable "Quartet".

I'm not saying that I'm going to get a miracle from one series, but that's the topic of the next thread, for now we just need to deal with one series.

Look at the file.

 

McDi is not a good choice.

If in terms of DSP, TA indicators are filters. Macdi is made based on MA, performance as a filter is poor. Better to make a makdi based on JJMA.

P.C.. Did a Fourier series decomposition based on Herzl algorithm (only one frequency component is calculated), so this component has amplitude and phase changing all the time.

 
Rorschach:

McDi is not a good choice.

If in terms of DSP, TA indicators are filters. Macdi is made based on MA, performance as a filter is poor. Better to make a makdi based on JJMA.

P.C.. Did a Fourier series decomposition based on Herzl algorithm (only one frequency component is calculated), so this component has amplitude and phase changing all the time.


how does it change? when fast, when slower, when galloping, it is interesting to see visually in front of the eyes to see this decomposition and its changes.

How the amplitude changes separately, how the phase changes separately, how changes in amplitude affect changes in phase and vice versa as the period increases, etc.

+ If Hertzel's algorithm is a decomposition for one frequency, then the other frequencies will also be interesting.

https://www.mql5.com/ru/forum/114579/page6#141176

 

Rorschach:

I did a Fourier series decomposition based on the Hertzl algorithm (only one frequency component is calculated), and this component's amplitude and phase keep changing.

Well, if the BP is not periodic or the period is not chosen correctly, it is bound to change.