Adaptive digital filters - page 24

New comment
 
faa1947 >>:Позиция на рынке - входим, выходи, вне рынка. Если мы говорим о сигнале, например, теле, то это изображение, которое может быть зашумлено. Если мы говорим о сигнале в геофизике, то до получения ВР делается модель сигнала (залежи полезных ископаемых), и потом пытаются его найти. По моим представлением сигнал на рынкете - это позиция

It's a bit of a mixed bag, but if this terminology, objects and their relationships are comfortable for you, then there's no problem, it's just a matter of definitions and "applications". The only thing is that - "a model of the signal (mineral deposit) is made before obtaining BP" is a bit wrong. There is knowledge about the parameters of the reflected signal change from some "formations/materials/...". The reflected signals are used to reconstruct subsurface structure. But sometimes one "listens" to subsoil, for instance to volcanoes and recovers with an eye that the signal is as complex as a quote :o) (if I've made up a little bit of it). Or, they just stick their "brains" into it and record it... But it's not sine waves there, it's very complex information, with a very complex nature.

What I wrote earlier was the Berg power density spectrum.

It wasn't easy to find out. :о) Not a big DSP expert, but I think the power density spectrum is a fundamental concept, derived earlier. And Berg suggested one of the parametric methods of estimating this quantity for autoregressive signal models.

 
bliznec1986 >>:
можно глянуть

Not yet, but maybe I will.

 
joo >>:

Сетями "сродство" выявлять тоже пробовали? Задача классификации вроде как.

Of course my IMHO: NS and classifications do not provide anything for this particular problem. The geometry affinity can be found by correlation and it will be easier and more effective than the most sophisticated NS :o) But I needed "process affinity" and therefore I needed to search for something that the process forms - a transfer function. And this is connected with identification and not so easy (for me anyway).
 
Farnsworth >>:

В общей формулировке, сигнал - это информация и совершенно не важно, источник искусственный или природный. Вы найдете огромное количество сигналов (именно сигналов) о происхождении которых ничего толком неизвестно (радиофизика, асторофизика, геология, ядерная физика, биология .................). Есть даже официальный раздел в ЦОС - "обнаружение сигналов", кстати, результаты этих исследований используются со всей серьезностью для обнаружения "осмысленных сигналов" в космосе :о) Есть раздел "случайные сигналы"

This is not entirely correct. Yes, the signals can be anything, but the applicability of a particular filtering method is based on the assumption of a well-defined model of the desired signal. In the case of a noise (random) signal as well.

 
Here's what I don't understand. Adaptive filters. Sounds nice. What do they adapt to?
Don't say "price." That's like saying nothing. What property of price should drive the adaptation?
Adapt what exactly? And what makes you so sure that your TS (and in my opinion it's not a very sensible thing to develop processing algorithms without considering a particular TS)
This adaptivity will improve something.

Otherwise, as usual, it's a conversation of simply nothing.
Otherwise, to paraphrase a famous Dumas character "I fight... because I fight" you could say "I filter because I filter".
 
For example, if you draw a sinusoid from the price, the adaptive sinusoid changes differently with every bar, its phase, frequency and amplitude change both ways and in different ways and from this different methods can also be used, for example, to monitor the speed and nature of phase or frequency changes and to find any dependencies between these changes and market movements
 
bliznec1986 >>:
к примеру если сторить синусоиды от цены то адаптивная синусоида с каждым баром меняется по разному меняется ее фаза ччастота и амплитуда и так и сяк совершенно поразному и от этого разными способами можно тоже плясать например следить за скоростью и характером изменения фазы или там частоты и находить какие либо зависимости этих изменений и движений рынка

Fascinating!

What's the point?
And why is it all down to sine waves?
I like exponentials better.

 
Well, for me personally it's easier with sine waves as it is easier to imagine them in my head, I can imagine how they change (adapt) because I am a programmer, so it happens
 
bliznec1986 >>:
к примеру если сторить синусоиды от цены то адаптивная синусоида с каждым баром меняется по разному меняется ее фаза ччастота и амплитуда и так и сяк совершенно поразному и от этого разными способами можно тоже плясать например следить за скоростью и характером изменения фазы или там частоты и находить какие либо зависимости этих изменений и движений рынка


How do you find dependency?
 
well, in general, the idea was presented as already wrote bliznec1986 wrote (a) >>
If we limit the harmonics to fit the price chart, a 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.

or on different timeframes it is possible to look for correlation of phases and their changes

I'm speaking in layman's terms (even if I have a university degree but not in these areas). I have ideas but I have to be a mathematician and a programmer to implement all that's why I have to use what's posted and search for interdependencies between them and check if someone has a ready-made adaptive filter (it may be based on different algorithms).

Programming is so shitty that even if I make a profitable system I have to stare into the monitor and use my hands, since I can't even experiment with inputs using ready-made signals .....

i cannot afford to order (i am so far poor), and even if i can pay for the work, i am not sure that the target will be the one i have been waiting for.....

it turns out that in order to test an idea a programmer needs n days, and me 1000n (that's an exaggeration), because I do all the projections in my head and have to think about it all the time, or eventually thoughts slip away, or try to jump to something else. then try to come back, but part of the thought has gone away and ...... and the computer is slowing the bastard (
1...171819202122232425262728293031...41
New comment