1st and 2nd derivatives of the MACD - page 6

 
nikelodeon:
You won't get anything good out of it, even if you take triple or decimal derivative. Yes, the first and the second derivatives sometimes show the top very quickly (without delays), but it is not long-term. Generally, at one time I twisted this chip for a long time, came to the conclusion that it is useless.....


What is useful?

I hope you don't think I'm trying to get any benefit by analyzing only one price with these methods (group movement, cluster, ..... that's what I mean).

 
nikelodeon:
You will not receive anything useful from it, even take triple or decimal derivative. Yes the first and second derivatives sometimes show the top very quickly (without delays), but it is not long-term. In general at one time long twisted this chip, came to a conclusion that it is useless.....


That's right, that's what I mean...

If we are going to make a physical analogy with the market, we should consider that in addition to speed, acceleration, and forces of inertia, there are also analogous forces of friction, elasticity, and gravitation (in general, forces that counteract the initial price impulses). By the way, I think Bill Williams wrote about something similar (the balance line is like a mountain top that is difficult to approach, but easy to move away from).

In general, physical analogies are certainly appropriate, but it seems to me that price movement is more like a trajectory described by the tip of a flexible car antenna than by a bouncing ball. And instead of understanding the intentions of the driver (the big market operators), many are trying to develop a statistically reliable "antenna tip movement theory" )))

 

And how are you going to calculate the derivative?

This is such a serious question that you could write a dissertation or two on it.

I mean "a good reliable derivative of a noisy signal".

Here the Russian guy Goloborodko, who lives in Japan and works for Olympus and others (I understand Japanese have problems with numerical methods), briefly describes the problem of FAST derivative computation in REAL conditions:

http://www.holoborodko.com/pavel/numerical-methods/numerical-derivative/smooth-low-noise-differentiators/

I guess he himself did not expect hundreds of comments from all over the world from specialists in various fields. His simple-minded surprise at the request to show the coefficients of the 3rd derivative is particularly hilarious. For all his skill, he has no idea who needs the 3rd derivative. Well ... He does not yet know that some particularly advanced numerical methods require good (reliable) derivatives of order 8...10 and no smooth splines help there.

So how do we calculate the differential?

 
trol222:


What's useful? Enlighten me.

I hope you don't think I'm trying to squeeze value out of analyzing only one price with these methods.(group movement, cluster, ..... is what I mean)


It depends on how you trade, what you use in trading. I agree with you that taking let's say 1 or 2 derivatives and putting it on D1, then you can get, say, 2-3 signals that will work for a month. This is D1. But when the signals start working and whether the next trigger is correct, this is an insoluble problem. And so.... A makd with its 1 or 2 derivatives is too narrow a tool, you cannot squeeze a stable and long earning system out of it....
 
AlexEro:

And how are you going to calculate the derivative?

This is such a serious question that you could write a dissertation or two on it.

I mean "a good reliable derivative of a noisy signal".

Here the Russian guy Goloborodko, who lives in Japan and works there for Olympus and others (I understand Japanese have problems with numerical methods), briefly describes the problem of FAST derivative computation in REAL conditions:

http://www.holoborodko.com/pavel/numerical-methods/numerical-derivative/smooth-low-noise-differentiators/

I guess he himself did not expect hundreds of comments from all over the world from specialists in various fields. His simple-minded surprise at the request to show the coefficients of the 3rd derivative is especially hilarious. For all his skill, he has no idea who needs the 3rd derivative. Well ... He does not yet know that some particularly advanced numerical methods require good (reliable) derivatives of order 8...10 and no smooth splines help there.

So how do we calculate the differential?


The usual way of thinking is: the ratio of the difference of the increment of the function to the difference of the argument. I have done this operation many times and have seen that there is no decrease in volatility as the order of the derivative increases to stop at any variant of the description. This shows the extreme non-stationarity of the process and the predominance of the exponential component in the hypothetical price formula. It is well known that the exponent is not "extinguished" by differentiation as other most functions are.
 
nikelodeon:

It depends on how you trade, what you use in trading. I agree with you that if you take let's say 1 or 2 derivatives and put them on D1, then you can get, say, 2-3 signals that will work for a month. This is D1. But when the signals start working and whether the next trigger is correct, this is an insoluble problem. And so.... A mcd with 1 or 2 derivatives is too narrow a tool, it is impossible to squeeze a stable long earning system out of it....


I haven't even brought up the system in one of my statements, what is important is the principle - the method of leading the price to a preemptive mode (to express the start of a move in a line - gather this move that is spread in time and in the whole cluster, and show it, at least approximately, for visual analysis)

And there are plenty of systems you can make from the data you build.

 
Zhunko:

This is a spectrum acceleration processed in a certain way. A 2nd order Chebyshev filter has been applied.

Does anyone here see a pattern?


They also tell me I'm talking in riddles.

Vadim, we are all very happy for your pictures, but for us mere mortals, they are like opium for the people)))) the point in showing them for themselves personally do not see, because I do not know what data they were built. this is like showing an orange circle from a distance and say it is an orange, and people will look for real oranges on those signs that have seen the orange circle (not knowing what should actually be orange).

The essence is not in the spectral analysis, but in the internal processes of the closed system, so there are some approximate frameworks for some processes. Simply if we do not replace the task of analysis of prices with the task of analysis of an oscillatory line we will not be able to bring various processes of pair behavior into one roughly speaking coordinate proscope.

I consider that it is not quite humane to make the increased emphasis on the minor things, not speaking about the main (it is better then not to speak about those or about others, if certainly it is a sincere desire to help, rather than a way to lead astray in selfish purposes).

 
Zhunko:

That's a strange answer. I didn't understand a word of it. Very abstruse.

The picture is clearly on topic. Read your topic title again.

The picture shows MACD acceleration. Only the filter there is not EMA or SMA, but Chebyshev 2nd order. The next derivative can be taken by eye. The slope of any line is the rate of acceleration. This is for those who don't understand.

Spectral analysis is the basis of any analysis. You can't do anything without it. Everyone uses it even without knowing about it. Those who just switch TFs also use spectrum analysis.

I have found a pattern in the picture that gives almost 100% positive inputs. That's to the point of how it can be applied. Surely that's what you're looking for? You don't just have an academic interest, do you?

Yes, I deviated from the topic a bit, sorry, I wanted to say that the essence is not in spectral analysis and not in filters and not in makdi (in this thread by the way I try to understand extrapolation process more deeply based on makdi), but when and to what data all this is applied - synthetic makdi at the output is needed, consisting of makdi of different tools of the cluster. You are posting pictures where lines are drawn using multicurrency analysis and people think (not me) that you can get such pictures using only 1 pair and say that the picture is clearly on topic.
 

I'll just drop the idea off at the moment so it doesn't get lost

https://forum.mql4.com/ru/38834/page288

 

Has anyone tried looking for macdi derivatives-which are based on renko, kagi or ranged bar charts? (how to extrapolate such macdies)

it would be interesting to build a makdi where the higher harmonic (sine wave-one complete cycle) consists of 2 lower harmonics (2 small cycles), and the higher chicle does not end until the 2 lower cycles have passed