Machine learning in trading: theory, models, practice and algo-trading - page 1925

 

Anyway, I screwed on a catbusta instead of trees, and... the problem was not in the trees, but as always, in the head :z

the method doesn't work as expected

 
Rorschach:
Nowhere did I come across any research on the best way to normalize the inputs: increments, subtraction ma, sliding window?

Or here

https://www.mql5.com/ru/articles/6351

Грокаем "память" рынка через дифференцирование и энтропийный анализ
Грокаем "память" рынка через дифференцирование и энтропийный анализ
  • www.mql5.com
Известно, что наличие большого количества участников на ликвидных рынках, работающих с различными инвестиционными горизонтами, продуцируют много рыночного шума. Другими словами, рынки обладают низким отношением сигнала к шуму. Ситуацию усугубляют попытки целочисленного дифференцирования временного ряда, которое стирает остатки памяти, приводя...
 
Maxim Dmitrievsky:

Anyway, I screwed on a catbusta instead of trees, and... the problem was not in the trees, but as always, in the head :z

the method doesn't work as expected

When did I ever tell you that? The problem is not you or the trees, it's the clusters, three clusters can't describe the +-exact entry point

That's why they have great acuracy, because everything is too coarse.

 
mytarmailS:

When did I ever tell you that? The problem is not you or the wood, it's the clusters, the three clusters can't describe the +-accurate entry point

That's why they have great acuracy, because everything is too coarse.

well even 2 describe on the lurn, on the new current doesn't work no matter what

and the error is small and everything seems fine... but the idea goes no further )
 
Maxim Dmitrievsky:

Let's think how we can generate signs in auto mode, accept the fact that we know nothing about the market, generate complete abstractions without any abstractions, it seems to me this is the only way ... So after millions of iterations we'll come to some kind of understanding

 
mytarmailS:

Let's think about how we can generate signs in auto mode, accepting the fact that we don't know anything about the market, that we can generate complete abstractions without any abstractions, it seems to me this is the only way ... So after millions of iterations we'll reach some kind of understanding.

I have a bot that goes through everything, it's useless.
 
Maxim Dmitrievsky:
I have a bot that goes through everything, useless

What everything? What does it go through?

 
mytarmailS:

What is everything? What does it go through?

chips and targets

did RL
 
Maxim Dmitrievsky:

chips and targets

did RL

does it go through or does it create?

 
mytarmailS:

he goes through or creates?

it both searches and creates, and searches from what it has created )) and creates from what it has searched ) and by mgua and all kinds of things )

and then selects the best ones through selection and rearrangement etc. etc.