Author's dialogue. Alexander Smirnov. - page 4
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
Here's a slightly idiotic model for you, Prival: if you consider the returns (signal increments), the signal is zero, the noise is a random process with a p.d.f. of the Cauchy distribution type and an ACF, which you know empirically. There are no measurement and quantization errors. Of course, the price as a result of integration will jump around the expected payoff, because the tails are very thick and dependent.
The model is extremely rigid, even tougher than the market, perhaps. But if your filter will work on such a model, it will work anywhere.
A very cool filter that fits your description is multiply by 0. If the signal is zero. :-). Works anywhere if you understand signal extraction as working.
LeoV
The question goes a little deeper. In order to answer that 1 indicator is more adaptive than another. You need to know what it should adapt to.
Adapts, of course, to the trend. The "bigger and stronger" the trend - the longer the JMA period. And that, as I understand it, is correct... .
Translate the concepts (trend, bigger, stronger) into the language of numbers. Then you can calculate and compare = say that this indicator is better than the other.
Prival, in Brownian motion, increments have an expectation of zero.
I know. And if the signal is 0. And the task of the TF is to isolate the signal, then the output of the optimal TF must be 0.
Not a concept, but a trend indicator, I wrote about it above (you probably did not read carefully) - the ADX indicator or a lawyer has a CFB indicator or.... well there are many of them.....
Not a concept, but a trend indicator, I wrote about it above (you probably do not read carefully) - the ADX indicator or, in jurik, the CFB indicator or.... well there are many of them.....
No, I read it carefully. Just trying to make it clear to you how I see it. That's the thing, that's your last sentence .... well there's a lot of them.... Where's the real one. The one, the best, the most adaptive, the one that best highlights the trend? The one, which has the property went down, I go down (drawdown=0), turned up, and I went up and again drawdown =0 and so on to infinity. And it works not retroactively as a zigzag, but at time moment t=0. (backwards can be built better than a zigzag)
Understand if we decide what is a useful signal that we have to filter=select=clear from noise. We have to know the signal and all its components + noise and its parameters.
Suppose
1. The signal is an equation of a straight line (directional movement = trend), many people have already made such a TF and written it in MQL4, it is available and freely accessible.
2. If it is a mix of oscillatory movements (sinusoids), it is another TF
3. if it is a mix of a straight line and oscillatory motions, it is the 3rd TF, etc.
If you define what the signal is then
This is a standard synthesis problem, there is input -> mixture of signal + noise, it is necessary to do (synthesize TF) on output of which by some criterion (better Bayesian) the signal is selected. For correct statement of the synthesis problem we need a mathematical description of the input.
If we take pictures that Djuric cites as a proof that his TF is better, more adaptive to mixture of sine and rectangular pulse(http://www.jurikres.com/catalog/ms_ama.htm#top),
Such signals are used in laboratory works in any normal radio engineering college. And there is mathematics and theory on how to filter them optimally for a long time.
To Mathemat
Returns kills the trend, that is what we can capitalize on. It has an ACF delta function it cannot be predicted. It's just noise that needs to be filtered out. What's left will be that clean signal we need, something we can capitalize on.
S.U. I've become a bad teacher, I can't explain everything in simple and accessible language, so that it would be clear what I'm talking about :-(
Ehhhhhhh.... if I'd have known I'd be living in Sochi....)))))))) I use CFB - and I'm happy with it. Although it's far from perfect.... as is the ADX....
Yes, you're welcome!
I ran a cursory glance at the article. I'm sure I just didn't understand the author!
In the place where it talks about the occurrence of group delay (GD) when using an anti-aliasing algorithm, the author offers a recipe for "getting rid" of the latter by using a reverse run. ... Is this a joke? It is known that for casual (working on the right-hand end of BP) systems, GZ is unavoidable in principle. But, of course, if BP is defined and we plan to work with it in the middle of a row (not on the right edge), we can, as the author advises, get rid of the lag by re-averaging with the same parameters in the reverse direction. But the author does not mention that, using such an averaging algorithm, we will inevitably see a re-drawing on the last bar. Has he forgotten about it or does he not know? Or what else?
Here is a quote from the article:
"Thus, with the above proposal we can partially compensate for the m/2 time lag (the first drawback of the traditional sliding average). And the second negative effect is eliminated ... And the third, and the fourth. ...
The use of the proposed averaging algorithm also significantly reduces linear frequency distortion... "
Neutron, thank you! Alexander, is the algorithm in Easy Language correct?
The algorithm in Easy Language is correct, but the programmer who implemented it has no experience of working in this language. The criterion of truth is practice. This algorithm, regardless of the size of the averaging window, provides a delay of 1 bar in the averaging product.