Machine learning in trading: theory, models, practice and algo-trading - page 1507

 
Alexander_K:

From the point of view of the theory of random processes, a trend is the sharpest deviation of the parameters of a random process from the same. A certain regularity appears - a directed, deterministic motion. These parameters are the so-called "discontinuity indicators" - entropy, Hurst, AFR, or anything else. You have to search, find it, and post it here.

Well, well. And you can't see it on the chart? No?
Look for the same deviations through... Um... Hearst... something. That's original.
ZZY I have doubts, but you understand in general, what and why you do.
 
Yuriy Asaulenko:
Well, well. And so, you can't see it on the chart? No?
The same abnormalities search through ... Um...Dick...Something. That's original.
I wonder if you know what you're doing and why you're doing it.

I do, and I keep the state open. But you? :))

 
Evgeniy Chumakov:


In what direction was the signal itself?

I don't know. My TS just filtered it out and that's it.

 
Alexander_K:

I understand, and I keep the state open. How about you? :))

Money matters are my personal matters. And no one should be concerned, by definition.
You're free to think what you want.))
 
Alexander_K:

I understand, and I keep the state open. How about you? :))

 
Has anyone tried to train a neuron to open trades and a second neuron to close them? - at the same time, of course.
 
Andrey Dik:
If I wanted to trade on ZZ, I should have learned to open deals from the first neuron and to close them from the second neuron. - At the same time of course.

I am now going to grow the forest on closing, just in the Expert Advisor will be two models - on the opening and closing.

The question of marking targets is very relevant - so far I've taken stupidly from the minimum take to the last 5 bars after the break of ZZ. The problem is that the sampling is very unbalanced.

The problem is that the sampling is not balanced. My Expert Advisor is trending and therefore it does not take all of the ZZ segments.

 
Aleksey Vyazmikin:

I am now going to grow the forest on closing, just in the Expert Advisor will be two models - on the opening and closing.

The question of marking targets is very relevant - so far I've taken stupidly from the minimum take to the last 5 bars after the break of ZZ. The problem is that the sampling is very unbalanced.

My Expert Advisor is trending, that's why it does not take all similar segments on ZZ.

If this is the case, I want to know if it makes sense at all and what success has been obtained (no need for statements, the word is enough), because this is the first time I'm experimenting in this direction.

 
Andrey Dik:

Ok, it will be interesting to know if it makes sense in principle and the progress made (no need for the stats, the word is enough), because I myself started experimenting in this direction for the first time.

And so far I haven't even trained the model, just made it possible to collect data.

Maybe later I'll share the results, but again, this will only be a story....

 
Ilya Antipin:

How is advisor Ilya?