create an expert for mt4 using a programme made in exel - page 25
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 know - I would also wish you better luck.
I personally have experience in predicting real processes after extracting significant harmonics.
And your failures are not a basis for hasty conclusions.
;)
Conclusions are made not on the basis of failures, and on the analysis of bases of method of decomposition in a Fourier series. This decomposition has limitations: it can only represent a function, periodic, on the decomposition segment. Accordingly, if a Fourier expansion is used, the function is assumed to be periodic, strictly f(x) = f(x+T), where T is the period. I hope you do not need to tell what value of the function you get when you try to extrapolate beyond the expansion segment for a periodic function? If done correctly and taken an infinite number of harmonics, then the corresponding from the beginning of the interval. If a finite number of harmonics, then accurate to the approximation error. The OOS is simply selecting the appropriate values from the beginning of the decomposition range ;) .....
Good luck.
ZY about real processes: they are predicted if there is a periodic component, e.g. cyclic load or carrier frequency, which, IMHO, we don't see in the market. The method itself is quite popular not only in radio engineering, but was popular in mechanics - it's easy to count integrals by hand (I counted in my time ;) ), with development of methods of numerical integration for mechanics the relevance is reduced......
The conclusions are not based on failures, but on an analysis of the basics of the Fourier series expansion method. This expansion has a limitation: it can only represent a function that is periodic on the expansion segment. Accordingly, if a Fourier expansion is used, the function is assumed to be periodic, strictly f(x) = f(x+T), where T is the period. I hope you don't need to be told what value of the function you get when you try to extrapolate beyond the expansion segment for a periodic function ?
Why do you think that the Fourier representation of a signal cannot be used for something more clever than to reverse transform and attribute the beginning of the signal to its end? Actually, that's the last thing on your mind. Your assertion, however, can be reworded roughly as follows: "everyone knows that two times two is four, so if someone's calculations include two times two, he's a fool, because whatever you do with it afterwards, you'll still get four." Sounds a bit silly, you must admit. If your study of Fourier analysis has not progressed beyond what you yourself have just described, I can only sympathise with you.
Why do you think that the Fourier representation of a signal cannot be used for something more clever than to reverse transform and attribute the beginning of the signal to its end? Actually, that's the last thing on your mind. Your assertion, however, can be reworded roughly as follows: "everyone knows that two times two is four, so if someone's calculations include two times two, he's a fool, because whatever you do with it afterwards, you'll still get four." Sounds a bit silly, you must admit. If your study of Fourier analysis has not progressed beyond what you yourself have just described, I can only sympathise with you.
Completely frivolous interpretation ;) - About the identity. As for 2x2 - can you give me an example where you can get something other than 4 by identity transformations?
If your study of Fourier analysis has gone so far that you can no longer see the limits of the method's applicability, I can sympathize with you in turn ;)...
Good luck.
Why don't we just get it out and measure it? We're professionals! :)
© AK
Completely frivolous interpretation ;) - About the identity. And regarding 2x2 - can you give an example where by identity transformations you can get something other than 4 ?
Who is talking about identical transformations? And speaking of bounds - who told you that the Fourier transform cannot be applied to non-periodic functions?
You can - but then the function is assumed to have a period equal to the size of the decomposition interval. In other words, it is still the values from the beginning of the range. I'm talking about the physical/geometric sense of the method. No tricks of Fourier decomposition method can be used for extrapolation, it's not meant for this purpose, that's all .....
2 -Aleksey- : I agree - I answered incorrectly, in the spirit of provocation. 2 alsu - My apologies......
Good luck.
And no Fourier decomposition method can be used for extrapolation - well, it's not designed for that, that's all ....
25 again.
You gave us an example with isolation and validation of significant harmonics - what prevents us from using them for short-term forecasting of non-periodic processes? We do not consider the signal to be a periodic function, nor do we consider its spectrum stationary at all, but we do imply that it contains certain harmonics with an amplitude varying slowly enough to solve the prediction problem for several counts ahead. Or do you think that Fourier will not work here as well?
25 again.
You gave us the example of isolating and validating significant harmonics - what prevents us from using them for short-term forecasting of non-periodic processes? We do not consider a signal to be a periodic function, nor do we consider its spectrum to be stationary at all, but we assume that it contains certain harmonics with an amplitude changing slowly enough to solve the prediction problem for several counts ahead. Or do you think that Fourier will not work here as well?
Frankly speaking, I don't think it will. For a long time I think so, I still could not formulate my feelings. Vladislav summed up my vague thoughts very clearly. Right in the hole.
// 2 VladislavVG by the way, thank you!
Will, apparently, popularise the gamma function and the corresponding probability distribution :)
I think the probability distribution is a long way off...
Although there is nothing wrong with jogging 'concepts'.
Will it add weight to the new forum?
Popularity, maybe.
;)