Machine learning in trading: theory, models, practice and algo-trading - page 2100

 
Maxim Dmitrievsky:

write an article

I can't do it, they'll laugh at me.

Aleksey Vyazmikin:

Very interesting, thank you for sharing. As soon as I figure out my current tasks, I'll try to run the code.

I don't know why you need it, the point is that there is a way to synthesize it, that's all, that's the whole point of the post

 
mytarmailS:

I can't help it, they'll laugh at me

Why do you need it, the point is that there is a way of synthesis and that's all, that's the whole point of the post

I am interested in the method, or you do not want that I do not use it?

 
mytarmailS:

I can't help it, they'll laugh at me.

Write it down, we'll laugh... only in details

because it's not clear on the forum. I did not understand them.

We need more topics about MoD in articles, but it's completely silent.
 
Maxim Dmitrievsky:

Yes, write it down, we'll laugh... but in detail

because it's not clear on the forum. I didn't understand them.

We need more threads about ONNX in the articles.

Well, write more. I'll even suggest a topic - the automation of machine learning. This is what almost all the members of this thread need.

I have something to write, but because of the reluctance of developers to see the codes in the R articles, I will not.

Concerning ONNX. Do you really need it? What can't you do without it now?

 
Vladimir Perervenko:

So write more. I have something to write, but because of the unwillingness of developers to see the codes to R in the articles - I will not.

About ONNX. Do you really need it? What can you do without it now?

I am preparing the sequel about data preparation for catbust.

To transfer deep learning to metaque, for example models in PyTorch

As for automation, I have it automated as much as possible. I get models at the touch of a button, which are saved immediately in the form of ready bots
 
Maxim Dmitrievsky:

I'm preparing a sequel about data preparation for catbust

Deep learning to be ported to metac, e.g. models on PyTorch

And without porting, does the model work (worse, slower ...)?

I see the need for ONNX for two categories of programmers - selling and hosting on VPS. And I don't see the profit from such applications.

The main task now is working models and approaches to create them. All the rest is a distant future.

 
Vladimir Perervenko:

And without porting does the model work (worse, slower ...)?

I see the need for ONNX for two categories of programmers - selling and hosting on VPS. And I don't understand the profit of such applications.

The main task now is working models and approaches to create them. Everything else is in the distant future.

It's harder to produce without porting, yes. It's harder to port to an oversite, it's harder to test on different accounts. And someone will want to sell on the local marketplace more

In MT5 it is much easier to "tune" the model by optimizing additional things like stoplosses, trailing bars, etc.
 
Maxim Dmitrievsky:

Without porting, it's harder to produce, yes. It's harder to migrate to an oversite, it's harder to test on different accounts. And someone will want to sell on a local marketplace.

I think it's easier to "tune" the model in MT5 by optimizing additional things like stoplosses, trailing bars, etc.

But the model won't work in the tester. No?

 
Vladimir Perervenko:

But the model will not work in the tester. No?

Of course it will, that's what I said, otherwise there is not much sense

 
Vladimir Perervenko:

Hmm, a thought occurred to me, is it possible to run the Python package for Metatrader5 in R-ka and use all the charms of Metatrader?