The market is a controlled dynamic system. - page 9

 
sergeyas:

What prevents you from also using a formula with adaptation depending on the current situation?

Tracking system == adaptation? Is there a fundamental difference?

The main thing in both systems is to find a "true" parameter for adaptation (read - some kind of misalignment signal in the tracking system).

It is true that a tracking system can also be "active", but an adaptive system is rather passive...

It's not the same thing at all.

I can hardly explain all the subtleties and details in this forum. It is impossible to do without literature.

 
avtomat:
Yeah, how can I put it... there is some dissatisfaction, a desire to improve... the search for ways to improve leads to some kind of improvement... and that, in turn, helps to see ways of further modification... it's an iterative process...


in principle for the market - yes, you can take several control lines...
... In principle, in relation to the market - yes, you can take several control lines... and then, depending on the choice of the best one (under the best conditions at the moment), put it in the model (top line)...
but this has already been mentioned... it's all sorted out... not a bad tool for visualization...put different things in and watch... everything is clear...

 

this is the picture you get - forward and inverse conversions

 
avtomat:

this is the picture so far - forward and inverse transformations


Yeah... cool... better than mine... I dabbled a while ago myself... but so far the best I've gotten is this...

I've given up so far... I don't know enough maths, etc... (trying to predict a white line)

 
Vizard:


Yeah ... cool ... better than mine ... I dabbled a while ago myself ... but so far the best I've been able to get is this...

I've given up so far... I don't have the maths skills...

that's a normal output. Is this the result of a neural network? which window was used? And I understand that an AR model is used?
 
avtomat:
that's a normal output. Is this the result of a neural network? which window was used? And I take it the AR model was used?

No... both without a window and without a grid... if you use a grid to predict a smoothed line, it's almost 1 to 1 and the grid output will be the same smooth (if by 1-2 bars)... but since it was shifted forward relative to the price quite a bit, the practical use is minimal (+ sensitivity to the quality of quotes)......your approach looks more interesting... Lately I've been doing (grid, regression) with model inputs that describe possible reversal moments-pre-conditions (for 1 BP)... But so))) not so much for practice... in general I dabble sometimes...
 

that's the kind of picture I'm getting... while I'm missing something...

.

three choices in the bottom three windows.

.

 
I'd like to see some indicators, pictures are pictures
 
avtomat:

that's the kind of picture you get... while I'm missing something...

It's strange. The pictures do not correspond to the ideology described earlier. You have 3 oscillators and there should be correspondingly 3 controls. Each of them switches on (off) the autogenerator at a certain time. Logically, this control is discrete, 1 works, 0 doesn't. Please describe pictures of what you have drawn here (built) mathematics (if possible) how one option differs from the other?

It is unclear what is here and how you get. That would be clearer, what you need to set forth, I will lay out my picture. With the assertion that my control is better (lower chart)

 
Trolls:

It's kind of weird. The pictures do not correspond to the ideology described earlier. You have 3 oscillators and there should be correspondingly 3 controls. each one works at its own time and switches the autogenerator on (off). logically this control is discrete 1 works 0 does not. Please describe pictures of what you have drawn here (built) mathematics (if possible) how one option differs from the other?

It is unclear what is here and how you get. That would be clearer, what you need to set forth, I will lay out my picture. With the assertion that my control is better (bottom graph)


No... It's just a very preliminary sketch and it's only the first part - movement in the opposite direction, and there's still a long way to go to the control, it's simply not there.

The problem is with the PF setting - when I introduce auto-tuning it will become clearer, because I myself am not quite sure yet where I should shoot...

However, if you are only interested in the final result, I will refrain from further demonstration of the search.