A purely theoretical question for mathematicians. With the possibility of moving to the practical plane. - page 8

New comment
 
And forecasting can be organised differently than presented in this gif, so this gif doesn't prove anything at all. But it well demonstrates the recovery of a function based on the raw data, which is extended to the next interval.
 
Nikolai Semko:

So what's the use. It has no effect on the quality of the forecast.

In the animated gif, the red line is the forecast line

you can also see it here:

https://www.mql5.com/ru/forum/216298/page5#comment_6484839

nice...just don't you think the function should be "mirrored".

It's just that even logically, from the symmetry of trades. One leverage deals in the past and the other, which is in the future, is in some way symmetrical to it. Whoever bought at the bottom will sell somewhere at the top, contributing to a slowdown and reversal. And if the price went straight down, he will get out earlier. Not exactly, but the central symmetry should show up.

 
Maxim Kuznetsov:

beautiful...just don't you think the function should be "mirrored".

It's just that even logically, from the symmetry of trades. One leverage trades in the past and the other, which is in the future, is symmetrical to it in some way. Whoever bought at the bottom will sell somewhere at the top, contributing to a slowdown and reversal. And if the price went straight down, he will get out earlier. Not exactly, but the central symmetry should show up.

It is simply the Fourier transform that decomposes the signal (price chart) for a given period into the sum of N harmonics (N sinusoids of different aplitude, frequency and phase shift). This is essentially what Kotelnikov's theorem tells us about.
Mirror it or not, makes no difference. The practical application for prediction is zero, just like flipping a coin. That's what the animated visualisation and code is for.
This is just one of an infinite number of possible finding patterns of numerical series. It's very naive to believe that you can find some universal pattern that can ultimately help predict future prices.
All coincidences are random and only temporary. That is my premise. I have already said it here. And this thread is useless mental gum, which just takes time and misleads immature minds.

 
Nikolai Semko:

The Fourier transform of a signal (price chart) for a given period is simply the sum of N harmonics (N sinusoids of different amplitude, frequency and phase shift). This is essentially what Kotelnikov's theorem tells us about.
Mirror it or not, makes no difference. The practical application for prediction is zero, just like flipping a coin. That's what the animated visualisation and code is for.
This is just one of an infinite number of possible finding patterns of numerical series. It is very naive to believe that you can find some universal pattern that can ultimately help predict future prices.
All coincidences are random and only temporary. That is my premise. I have already said it here. And this thread is useless mental chewing gum that just takes time and misleads immature minds.

There's just a nuance - we don't have an abstract number series. There is a causal relationship between the past expressed in the chart and the future. These are open trades. We don't know exactly when and to what extent they will be realized, but we can suppose that it will cause some symmetry. The symmetry will most likely appear after the "impulses" - the price has suddenly gone upwards, those who were in shorts has run away on stops or has left. The former ones remain in longs, and their closes will be quite symmetrical (blurred, but not the point).

In extrapolating, we forget about the second arm. We extrapolate the first one, which has already happened, from other circumstances and will not happen again. We may have an approximate correct result, we interpret it incorrectly, the sign should be different. By analogy, like energy, which is composed of potential energy and kinetic energy.

 
Yes, I forgot all about Fourier, it does an excellent job of interpolation. I think a lot of people have dabbled with it. I have problems with extrapolation though, the results are strange, sometimes it hits, sometimes it doesn't.
But what's interesting is that if you try to predict with a Weirstrasse Fourier function, you get a finger in the sky forecast (if you take less than a period for analysis). Although the fuction is not random and has a well-defined formula.
Probably any method of analysis will lead to similar results unless there is an exact formula specific to the series.
 
Aleksey Panfilov:

Yes.

Here's a Fourier near an inclined straight line, for example. ))

Fourier near a square parabola:


Or a polynomial:

What's wrong with calculating coefficients?

The series shown in the first post is purely abstract.

The real series looks like this:

-1,202222222 0,378626831 -0,15119 0,249393 -0,31012 0,268994 -7,48 0,169875 -0,04285 0,19273 -0,03837 0,040593 -0,14868 0,059058951

or as follows

-36,06666667 22,33898305 -16,1776 23,19355 -21,0882 19,09859 -112,2 21,06452 -10,9688 20,81481 -10,3593 8,686916 -19,0313 12,69767442
 
Yousufkhodja Sultonov:

Is there any positive result on the demo?

 
Сергей Таболин:

The series given in the first post is purely abstract.

The real series looks like this:

-1,202222222 0,378626831 -0,15119 0,249393 -0,31012 0,268994 -7,48 0,169875 -0,04285 0,19273 -0,03837 0,040593 -0,14868 0,059058951

or as follows

-36,06666667 22,33898305 -16,1776 23,19355 -21,0882 19,09859 -112,2 21,06452 -10,9688 20,81481 -10,3593 8,686916 -19,0313 12,69767442

Based onhttps://www.mql5.com/ru/code/10339



Индикатор Султонова
Индикатор Султонова
  • www.mql5.com
Индикатор состоит из трех линий- селл(красная), бай(синяя) и линии трейдера (желтая). Он прогнозирует предполагаемый ход цены в будущем, анализируя заложенную историю в виде заданной ретроспективы. При осуществлении торговых сделок нужно придерживаться желтой линии трейдера, являющейся преимущественной линией движения цены, которая указывает...
 
Renat Akhtyamov:

Do you have any positive results on the demo?

I will start cents real soon and put up a signal.

 
Yousufkhodja Sultonov:

I'll start a cent real soon and put out a signal.

Ahh...

By the way, to put it mildly, the internet is littered with


12345678910111213
New comment
close