Machine learning in trading: theory, models, practice and algo-trading - page 2694

[Roman]

Renat Fatkhullin #:

This is proof that third-party solutions are not suitable - people need regular MQL5 applications.

In this context, there is a proposal.
There is a lack of standard functions for building scientific graphs.

Sometimes you need to quickly display data or a calculation for visual characterisation.
And you have to go to the library and figure out how to display scientific graphs.
Which takes time and attention away from the main task.

[mytarmailS]

Roman #:

In this context, there is a proposal.
There are not enough standard functions for building scientific graphs.

Sometimes you need to quickly display data, or a calculation for visual characterisation.
And you have to go to the library and figure out how to display scientific graphs.
Which takes time and attention away from the main task.

It will be endless, no matter what they add, there will always be something missing... You will always want something new and convenient as in other languages, the library base of which is constantly growing, every day, every hour.....

But you can't use it, it's a crutch, it's duct tape, write it yourself if you need to....

Here came the thought, isn't mcl code in mt itself a tape? Isn't it a crutch in the context of modern requirements?

I wonder...

Valery and I were writing one TS, he wrote it on mt5, I wrote the same thing on R, I got ~30 lines of TS code+contact to mt5+overrides. Errors, and he has more than 150 lines of code on pure mt5, how is it at all?? This programme is designed for this and it is so miserable, how? ...?

[Roman]

mytarmailS #:
It will be endless, no matter what they add, there will always be something missing... You will always want something new and convenient as in other languages, the library base of which is constantly growing, every day, every hour.....

But you can't use it, it's crutches, duct tape, write it yourself if you need to....

Here came the thought, isn't mcl code in mt itself a tape? Isn't it a crutch in the context of modern requirements?

I wonder...

If MQ policy has started to think about end-user convenience,
then correct suggestions should be made to implement these conveniences.

[mytarmailS]

Roman #:

If the MQ policy started thinking about end user convenience,
then correct suggestions need to be made to implement those conveniences.

They are not thinking about the user, they are thinking about the sellers from the mockup and the lock-in to mql

If they were thinking about users, there would be an api for the platform

[Evgeny Dyuka]

mytarmailS #:

I was wondering if the µl code in mt is not a crutch in the context of modern requirements?

Good question. It is certainly not duct tape, but it is definitely an outdated branch of development.
Nokia also dominated for 20 years, then did not notice the wind of change and disappeared very quickly.

[mytarmailS]

Evgeny Dyuka #:

Good question. It is certainly not duct tape, but it is definitely an outdated branch of development.
Nokia also dominated for 20 years, then did not notice the wind of change and disappeared very quickly.

Well personally for me/for my needs it's a crutch that doesn't fulfil my needs

[Renat Fatkhullin]

Roman #:

In this context, there is a suggestion.
There are not enough standard functions for building scientific graphs.

Sometimes you need to quickly display data or calculations for visual characterisation.
And you have to go to the library and figure out how to display scientific graphs.
Which takes time and attention away from the main task.

Visualise it! Graphics library in MQL5 as an analogue of plot from R

This has been a standard solution for 5 years.

[Renat Fatkhullin]

mytarmailS #:
They don't think about the user, they think about the vendors from the mockup and the closed mindedness of the µl

If they were thinking about users, there would be a platform api.

99% of users are users. And they are the ones who download apps from the market, use signals and codobase.

It is for them that we have created many services.

And for those who can write professional programmes, we are developing the language. For this purpose, more and more functionality is being added to MQL5.

The next big step is the development of machine learning using in-house tools. And you will almost never have to retrain from Python.

Even now, many matrix operations are close to the similarity with Python packages like Numpy.

[Roman]

Renat Fatkhullin #:

Visualise it! Graphical library in MQL5 as an analogue of plot from R

This is a standard solution for 5 years already.

Renat, we know very well about the library.
Why not wrap it in MQL standard functions?

Plot(...);
ScatterPlot(...);
HistogramPlot(...);

etc.

This was my suggestion.

Forum on trading, automated trading systems and testing trading strategies

Machine learning in trading: theory, practice, trading and more

Renat Fatkhullin, 2022.08.20 07:27 PM

This is proof that third-party solutions are not suitable - people need in-house MQL5 applications.

[Evgeny Dyuka]

Renat Fatkhullin #:

Taking into account your words about "complicated language, I don't understand", you don't have serious knowledge, outlook in this topic and understanding of behavioural strategies of companies-developers-calculation-platforms.

The problem with the topic of neural networks in trading is not in the tools, there are plenty of tools (I am not talking about mql). And there are plenty of professionals who"have serious knowledge and outlook in this topic".

On the basis of popular libraries this problem has been solved for many years. Maybe the problem is not in tools and professionals, but in ideas and understanding of HOW to do it. You can always get professionals to a working solution and they will polish it.

My point is that your race for beauty of tools may not lead to results. Other places have these tools for a long time, as well as access to historical quotes data. You are not giving anything new. Why is it suddenly going to work here?