Machine learning in trading: theory, models, practice and algo-trading - page 3133

 
Ivan Butko #:

Thank you

https://www.mql5.com/ru/articles/8863

there in the search text:

"The X axis shows the frequency of trades, or rather their duration, in bars. Y-axis shows the estimation of R^2 for each of the passes. It is well seen that too short trades in 0-5 bars badly affect the performance of the model, while the range of 15-23 is optimal. Increasing the duration of trades above 30 bars starts to degrade the results. There is a small cluster with transaction durations of 6-9 bars, for which the estimates are maximum. Let's try to train models with such durations and compare the results with other clusters".

Поиск сезонных закономерностей на валютном рынке с помощью алгоритма CatBoost
Поиск сезонных закономерностей на валютном рынке с помощью алгоритма CatBoost
  • www.mql5.com
В статье показана возможность создания моделей машинного обучения с временными фильтрами и раскрыта эффективность такого подхода. Теперь можно исключить человеческий фактор, просто сказав модели: "Хочу, чтобы ты торговала в определенный час определенного дня недели". А поиск закономерностей возложить на плечи алгоритма.
 
Forester #:
Train the AI to ban. Not only for foul language, but also for rudeness. On the basis of content analysis.
Then also for inconsistency with the subject of the branch (it is more complicated).

It's not that complicated anymore.

Here's a first attempt.

============


The api is open, take it and do it.

 
mytarmailS #:

It's not that complicated anymore.

Here's my first attempt.

============


the api is open , take it and do it

Idealiti:)
 
Maxim Dmitrievsky #:
Idealiti:)

I like it myself.)

And the branch would immediately lose 95% of its weight ...

who's interested in that?

 
mytarmailS #:

I recommend learning about genetic programming so that you don't have to do your research by rote.

In order to search for regularities not tied to time/chronology I recommend to get acquainted with algorithms asocial rules.


And the direction of the search is correct.

there are only prices and a sequence of certain events on the market.


is that if you open a point B at point A of exactly the same nature?

 
mytarmailS #:
What???

levels

 

what tools can be used to describe any regularities of any type?


1) they must be rules

2) functional transformations

3) code

4) mixtures of different methods

 
mytarmailS #:

what tools can be used to describe any pattern of any type?


1) they must be rules

2) functional transformations

3) code

4) mixtures of different methods

A regularity is a regularity because it is already described: there is a formula for it, for example, Newton's law of universal gravitation, or a whole theory.

And in general, tools are selected for the type of processes, which can be:

  • deterministic
  • random
  • uncertain, i.e. random, where a human being is a part of the process, e.g. financial markets.
 
mytarmailS #:

what tools can be used to describe any pattern of any type?


1) they must be rules

2) functional transformations

3) code

4) mixtures of different methods

repeatability that can be determined by something.

The environment plays a big role, and types like deterministic or random depend on the ability to account for/define something.

And in essence, getting rid of noise is the same response to the same impact)

 
In graphs, if you just describe it :)
Reason: