Help with Fourier - page 3

[Сергей]

Integer писал (а):

... maybe there is some benefit to it.

The author (I mean klot), as I understand it, has a hypothesis/assumption that ....mmm... let's say that there is a different "spectrum" in (different) market phases/states (or at least that some spectral parameters/characteristics change when moving from one state to another. However this should be checked first, in principle). At least this is what I have read from his comments to the posted code. (Everyone sees only what he wants to see! - I mean me, of course.)

[Юрий Макаров]

Fourier is not working in the market.
Wavelets are better, but they don't work either :)
As the contest shows, the crossover on the muwings works :))
(lucky people ...)

[Sergey]

Fourier may not work, but ... In nature everything has an oscillating process and this is an irrefutable fact. And the same laws apply to forex. Using this fact, you can look into the future :)

[Dmitrii]

SergNF писал (а):

Integer wrote (a):

... maybe there's some benefit to it.

The author (I mean klot), as I understand it, has a hypothesis/assumption that ....mmm... let's say that there is a different "spectrum" in (different) market phases/states (or at least some spectral parameters/characteristics change when moving from one state to another. But this should be checked, in principle, first). At least this is what I have read from his comments to the posted code. (Everyone sees only what he wants to see! - I mean me, of course.)

Different market states have really different spectrum, try to run my indicator in visualizer, which outputs spectrum in a separate window (you need to input aa[i]=Close[i];)

[Dmitrii]

The fact that the spectrum is "floating" is in principle obvious, and it has been mentioned more than once in various literature. I am now trying to classify market state spectra with the help ofa neural network, i.e. to conduct something like a cluster analysis.
If the neural network draws the right conclusion about whether the market is trending or flat, we can use different trading systems. All in all, this is for MTS-oks...

[Dmitry Fedoseev]

klot писал (а):

The fact that the spectrum is "floating" is in principle obvious, and it has been mentioned more than once in various literature. I am now trying to classify market state spectra with the help ofa neural network, i.e. to conduct something like a cluster analysis.
If the neural network draws the right conclusion about whether the market is trending or flat, we can use different trading systems. All in all, this is for MTS-oks...

Oh! If only it were permanent.

[Dmitry Fedoseev]

Here's my fakes on the subject, but I'm not sure if it's all right.

[Dmitry Fedoseev]

...

[Stan Podlisker]

Rosh:
Kotelnikov's theorem.

This is actually Nyquist's criterion.

[Юрий Макаров]

There's the Kotelnikov theorem, and there's the Nyquist frequency. :)
One does not prevent the other ...