Adaptive digital filters - page 23
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
If we limit the harmonics to fit the price chart, then the large harmonic with a large period can be decomposed into small harmonics (also limited by the variation) that are part of large ones and on their basis predict the movement of the same large harmonic, respectively within the limits where these limitations allow.
Each harmonic has its lifetime (it is a trend). The wavelets give the harmonics in frequency-time coordinates. You can see how these bumps emerge, grow and die. Maybe this is where your idea is implemented?Уточняю: гармоники есть всегда, только живут они обычно недолго и вообще время их жизни неопределенно. Нашел гармонику, а она взяла и померла или не померла. Мы какую обсуждаем, ту что померла или ту что не померла?
I completely agree with you. Harmonic decomposition for the quoting process does not carry any information. The point is not what died/survived, the point is that these harmonics are completely random and of no use. It only makes sense to consider the power spectrum, it carries information.
I am writing about DSP. Radio, TV, location etc. There is always a signal source and we have a certain, a priori knowledge of that signal.
In general terms, a signal is information and it does not matter whether the source is artificial or natural. You will find a huge number of signals (exactly signals) about the origin of which nothing is well known (radiophysics, astrophysics, geology, nuclear physics, biology .................). There is even a formal section in DSP - "signal detection", by the way, the results of these studies are used in all seriousness to detect "meaningful signals" in space :o) there's a section on "random signals"
there's a lot of it....
If the price is a signal, what is it about?
It says things like EURUSD etc. :о) So a signal about the state of the voltage in the socket doesn't confuse you, but a signal as a quote is something strange?
The price sequence, BP, should give us a market position signal. No separation of noise gives a market position, there are no algorithms, no maths for that. The classical approach is to identify BPs (get some parameters) and having identified BPs by algorithm get a position.
I don't really understand what "market position" means. If you mean the location of the "now"/"yesterday" quoting process, then all the information is there. if you mean trading decisions, then yes, you need to identify the process, and that's not so easy.
I have seen an article where it is proved that it is impossible to identify the trend as such.
It's not quite like that. I won't argue now, but will come back a bit later, but with an argument :o)
тут один человек вроде как я понял уже построил адаптивный фильтр на основе алгоритма Герцеля еще человек на основе ких если я не ошибаюсь а у каго еще получалось сделать адаптивный фильтр может на основе калмана или еще какие?????
I did once based on an optimal Wiener filter. The reference signal for each "now" was derived from a statistical analysis of "related" processes in history. Determining the "affinity" proved to be a non-trivial task.
Position in the market - in, out, out of the market. If we are talking about a signal, e.g. a body, it is an image which can be noisy. If we are talking about a signal in geophysics, a model of the signal (mineral deposits) is made before BP and then one tries to find it. The way I see it, the signal in the marketplace is a position.
I completely agree with you. Harmonic decomposition for the quoting process does not provide any information. The point is not what died/survived, the point is that these harmonics are completely random and are of no use. It only makes sense to consider the power spectrum, it carries information.
Everything I wrote earlier was referring to Berg's "power density spectrum".
Я сделал когда то на основе оптимального фильтра Винера. Эталонный сигнал для каждого "сейчас" получал на основе статистического анализа "родственных" процессов в истории. Определить "сродство" - задача оказалась нетривиальная.
Have you tried to identify "affinity" with networks as well? A classification task of sorts.