Machine learning in trading: theory, models, practice and algo-trading - page 3454

[Vladimir Perervenko]

mytarmailS #:

Well, that's what I'm asking why.

What can python lib for MT5 that mt5R can't?

What is the profit of pulling python into your R script (which is not needed there at all)?

If you don't use python packages, then of course you don't need python. I repeat, the post is about the interaction between MKL and Python. Does your variant solve this issue?

[Maxim Kuznetsov]

Vladimir Perervenko #:
Just for interest: where and how you specify Python version, environment from which to get packages, etc. HeloWord is good, but have you implemented anything more serious?

There's a $PYTHON_HOME that specifies where Python lives.

for me Python is a side-effect ;-) just is and is. defaults: current 3.7.7, not system - separately built.

seriously, I'm using Tcl from MQL to get event-queue,threads,shared vars,gui,redis. That is, to do what in MQL is not possible at all, but without it I can't go anywhere.

[Vladimir Perervenko]

Maxim Kuznetsov #:

there is a $PYTHON_HOME that specifies where Python lives.

for me Python is a side-effect ;-) just is and is. defaults: current 3.7.7, not system - separately built

seriously, I'm using Tcl from MQL to get event-queue,threads,shared vars,gui,redis. That is, to do what in MQL is not possible at all, but without it I can't go anywhere.

I see. I do all the same things in R. Python of different versions only for models.

[Maxim Dmitrievsky]

Aleksey Vyazmikin #:

After starting the terminal, 21 seconds

Repeated data request 13 seconds

And again - almost instantaneous 0.5 seconds.

In general, experiment - the code from the terminal will work

Do they write codes for you? A well-structured code doesn't look like a beginner's code :)

Or this is from the documentation

At first terminal was pulling history from server, cold query. Then it should be very fast.

[mytarmailS]

Maxim Dmitrievsky #:
Do they write code for you? Well structured code doesn't look like beginner code :)

Or is it from the documentation

At first the terminal was pulling up the history from the server, cold query. Then it should be very fast.

And there are no perennials with names like PEREMENNAJA_x1. :)

[Maxim Dmitrievsky]

mytarmailS #:
And there are no perennials with names like PEREMENNAJA_x1. :)

Yeah :)

[Aleksey Vyazmikin]

Maxim Dmitrievsky #:
Do they write code for you? Well-structured code doesn't look like beginner code :)

Or is it from the documentation

Most of the code is from MQL documentation. The function was modified to suit the needs from the article published here. ChatGPT offered a really useful solution in the form of a list of TFs. In general, I just used different sources and compiled into one whole, the main thing is to understand what you are looking for. I have a lot of experience in writing MQL code, so the question is to find similar functions and solutions.

Maxim Dmitrievsky #:
At first the terminal was pulling up the history from the server, a cold request. Then it should be very fast.

The history was already there, as I ran the code several times during debugging. Most likely there is some caching going on, i.e. the history for the requested period is unzipped for quick access during the terminal session.

[СанСаныч Фоменко]

Vladimir Perervenko #:
Using the mt-R library, which has been tested for many years. Globally, this makes it possible to use MQL/R/Python together both in combat and testing, with a small change. When testing quotes and commands through MQL and when working through Python.

Now it's clear how the python apologists have got to such a life that TS evaluation is done by p-square, and they just can't afford a tester!

[mytarmailS]

СанСаныч Фоменко #:

Now it is clear how python apologists have got to such a life that TC estimation is done by p-square, and they just can't afford a tester!

What does Vladimir's post have to do with it?

[Maxim Dmitrievsky]

https://www.mql5.com/ru/articles/2358