Help with Fourier - page 5

New comment
 
Rosh:
What is meant by prediction accuracy?

A projection is drawn into the future. If the actual trajectory then coincides with the calculated trajectory, then the accuracy is considered high. How accurate is accurate? Well, it is elementary to compare the two graphs and use them to make some statistics, if necessary.
 
New писал (а):

Well, the Fourier thing about knowing how to use Fourier is about the same as saying
is like saying of course tech analysis works, you just have to know how
how to use it.

Well, that's probably the most important thing. It takes much time and mental effort to master the understanding of the process. Besides, you need to be able to program well and do it all. Then, as in learning, say, a foreign language, after a certain threshold, something begins to become clear, the unnecessary things go away and simple basic rules remain.
 
ANG3110, please show me a screenshot of the whole analysis period, I need to see the beginning and the end of it, you have already posted the end, now I need to see the beginning.
 
To understand how this works, it's easiest to use music samples as an example. If the quality of the reverse loops is high, the sound is good, if it is low, it is bad. Once again, just as for musical samples, you take a few periods and leave the ones that repeat each other's wave patterns, and the others are considered to be alternate.
 
ANG3110:
Rosh:
What is meant by forecast accuracy?

A projection is drawn into the future. If the actual trajectory then coincides with the calculated trajectory, the accuracy is considered high. How accurate is accurate? Well, simply compare the two graphs and use them to make statistics if needed.
I see. The matter is that MTS based on linear regressions (the famous thread on another forum) shows that 67% of entries were profitable, but it does not give the desired profit, so the profit and moose are different in each case.
 
ANG3110, this is not a Fourier, most likely very heavily modified. If you had built a real Fourier, you would have the next point after the last analysed point, i.e. the 1st predicted point, coincide in value with the 1st analysed point. That is, the end would go to the beginning. This is a feature of the Fourier construction. Tell us how your indicator is constructed.
 
Rosh:
ANG3110 wrote (a):
Rosh:
What is meant by prediction accuracy?

A projection is drawn for the future. If the actual trajectory then coincides with the calculated trajectory, then the accuracy is considered high. How accurate is accurate? Well elementary compare the two graphs and you can do some statistics on them, if necessary.
I see. The matter is that MTS based on linear regressions (the famous thread on another forum) shows that 67% of entries were profitable, but it does not give the desired profit, so the profit and moose are different in each case.

...And the slope of the linear regression and width of the channel are also different ...Entries that cross the 60 percent RMS in one direction will not be equal to entries in the other, the shoulders are different. Is this taken into account?
I cited a particular case of linear regression without envelope by RMS. In practice I use a more complex construction that takes curvature into account.
 
lsv:
ANG3110, this is not a Fourier, most likely very heavily modified. If you had built a real Fourier, you would have the next point after the last analysed point, i.e. the 1st predicted point, coincide in value with the 1st analysed point. That is, the end would go to the beginning. This is a feature of the Fourier construction. Tell me how your indicator is built.
I may tell you about it later, if I have time and the feeling that it is needed. At the moment I have no cable Internet line, I am writing via phone modem, but it is expensive. So that's it for now.
 
I'm on GPRS at home and no problems :)
ANG3110, why did you remove the picture, I liked it. If you look closely at this picture, you can see the following, you have Fourier there, but as I understood, instead of the horizontal line the centre line of the channel is used. Now take a closer look, you have future values repeating the initial values of the analysed period. Am I right or not?
 
lsv wrote:
I'm on GPRS at home and no problems :)
ANG3110, why did you remove the picture, I liked it. If you look closely at this picture, you can see the following, you have Fourier there, but as I understood, instead of the horizontal line the centre line of the channel is used. Now take a closer look, you have future values repeating the initial values of the analysed period. Am I right or not?
That's right, I didn't mention music samples for nothing. The end cycles with the beginning. It is important to find the repeating periods out of several. The market fluctuations are of periodical nature. The opening times of the various exchanges are repetitive, news release times, weekly cycles, 2-day price reaction, seasonal fluctuations, the Earth is believed, periodic flows from the cosmos, solar activity, etc., etc. This creates a set of oscillations with different frequencies and when they interact with each other they create sub-harmonics. If you consider the slow oscillations and the fast ones superimposed on them, you get quite complex patterns. Here's a seasonal 1/3 year cycle for example. It is fairly constant. When the phase coincides there are jumps. The Euro deviated from the seasonal cycle three times, and when the phases coincided, we have what we have now - a strong market breakout, and before that the phases did not coincide. I purposely gave an example based on deviations from a linear regression, otherwise, if I had given an example with curvature, i.e. slow harmonics, and in addition the sum of several cycles, you would not have understood anything, but as you see, everything is simple and easy.
123456789101112...19
New comment