1st and 2nd derivatives of the MACD - page 38
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
You can take many narrowband filters instead of one broadband filter and not have to move anything.
There was a branch three years ago.
I am transferring two graphs from there.
The bursts are amplitude maxima at frequencies (or rather periods = inverse of frequency.
The algorithm itself, supposedly by Burg, is a maximum entropy filter. Available in Matlab.
Very nice graphs, but when you change the size of the window, and even worse when you shift the window, the look of the graph changes. I.e. those resonant frequencies are rigidly related to the window. Maximum entropy simply summed up the amplitudes at one frequency and that's it, and what it would be different with a shift, that question is not solved - so we can't use the information outside the graph.
Where did you get this spectrum from? I've shown a view of the spectrum using FFT - quite different ~ 1/f^2 without resonances.
Where did you get this spectrum from? I've shown a view of the spectrum using FFT - quite different ~ 1/f^2 without resonances.
Most likely what the person is trying to say is to use a system of filters, probably something to do with the delay bands of the filter, the delay bands are probably selected so that each successive filter complements (or continues) a higher order filter or something like that.
let's say one filter filters a certain frequency band, another filter filters another frequency band, and the third filter filters skips frequencies....... everything.... go.... not my topic...I will do it in a year, for now I will try to do without it, if it works out.
MATLAB 7.0
This is the AFC according to Burg
I have been trying to explain the essence of econometric models here for a long time. I will have to do it mathematically in this post after all. Burg is an autoregressive (AR) model:
x[n] = a[1]*x[n-1] + a[2]*x[n-2] + ... + a[P]*x[n-P]
Apply the Z-transformation and obtain the characteristic equation of this model
z^P = a[1]*z^(P-1) + a[2]*z^(P-2) + ... + a[P].
Solve this equation and find its complex roots Z[1] ... Z[P]. Each complex root is
Z[k] = Exp(q[k] + j*w[k]) = |Z[k]|*Exp(j*w[k])
where k = 1...P and j is an imaginary unit. If all |Z[k]|<1, then our AR model is stable. We rewrite our AR model as the sum of the roots of the characteristic equation:
x[n] = h[1]*Z[1]^n + h[2]*Z[2]^n + ... + h[P]*Z[P]^n
or
x[n] = h[1]*Exp(q[1]*n+j*w[1]*n) + h[2]*Exp(q[2]*n+j*w[2]*n) + ...
So the AR model, whether Burg, Yule-Walker, or Prony, tries to fit the damped oscillations into our series, where w[k] is the frequency of the oscillation. What you have shown in the graphs is not the spectrum of the quote, but the spectrum of the Burg model. And the positions of so-called "price resonances" reflect the position of the roots of this model on the frequency response. A change in prices leads to a change in the coefficients of the Burg model and a drift of our roots and "resonances".
All this econometrics boils down to regression. Talking about the physical meaning of the oscillating solutions of the AR model is as successful as talking about the physical meaning of the coefficients of a polynomial regression or the formula (18) of the Yusuf model. Take a regression function we like and talk about it for 300+ pages as an achievement of humanity.
Would it be possible to find a maximum entropy filter in Matlab and post the result of the application here.
Gentlemen, don't be lazy. Download the software and have a look. Better yet, believe that studying the simplest things is much more useful than maximum entropy, stochastic multifractal and spherical horse in a vacuum. I once dabbled in such complex science as wave analysis a long time ago. I gave it up when I realised the simple reason it would never work in forex.
Gentlemen, don't be lazy. Download the software and have a look. And better still, believe that studying the simplest things is much more useful than maximum entropy, stochastic multifractal and spherical horse in a vacuum. I once dabbled in such complex science as wave analysis a long time ago. Quit when I realised the simple reason it would never work in forex.
What simple reason?
It works sometimes - that's for sure.
What kind of bandpass filter is used, if not a secret? Built into MATLAB?
There's nothing built in. There's a utility to calculate whatever filters you want. Whatever you want, that's what you get.