Machine learning in trading: theory, models, practice and algo-trading - page 2393

[Maxim Dmitrievsky]

Evgeni Gavrilovi:

Got to the fit, then the error: fit() missing 1 required positional argument: 'timeseries_data'

Probably need some other format for feeding time series

I'll get to it later, I don't have time yet. I wanted to write my generative using lstm.

[Maxim Dmitrievsky]

Evgeni Gavrilovi:

Apparently you need some other format for the time series

https://sdv.dev/SDV/user_guides/timeseries/par.html

Yes, there is in the description

[Aleksey Vyazmikin]

Maybe someone has not seen, but I created a thread in which I ask for help in solving the problem. So those wishing to stretch their brains welcome.

[Evgeni Gavrilovi]

Maxim Dmitrievsky:

sent a new article for review, with new ideas

the article is not approved? something it does not appear in the list

[Maxim Dmitrievsky]

Evgeni Gavrilovi:

Didn't you approve the article? It doesn't appear on the list.

I decided to hold back, there were some changes, which have no time to finish

[Evgeni Gavrilovi]

Maxim Dmitrievsky:

I've decided to hold back, there are some things I have no time to finish yet

If you want to convert your EA to MQL4, should I rewrite everything? Or it can be done with a single includeMT4Orders?

[Maxim Dmitrievsky]

Evgeni Gavrilovi:

to convert your EA to mq4 do you need to rewrite everything? or it is done with a single includeMT4Orders?

you need to remove mt4Orders and see what other compilation errors you need to fix in your code, because the languages are a bit different

[Alexander Ivanov]

Hi!

Here's an article about neuro-prediction -https://otexts.com/fpp2/nnetar.html

Read it if you're interested. There are many examples, codes, etc.

[mytarmailS]

Alexander Ivanov:

Hi!

Here's an article about neuro-prediction -https://otexts.com/fpp2/nnetar.html

Read it if you're interested. There are lots of examples, codes, etc.

Nah, it doesn't work for the market...

First, only price is taken into account, and this is a single-domain vector... And what if you want to work with 3k signs?

The solutions are dimensional reduction methods.

Secondly, the market is NOT a time series, so in fact it is, but in fact it isn't, the market doesn't obey those laws of those algorithms that are made for time series prediction...

The solution is to look at the market from the point of view of levels or limit orders.

[Aleksey Nikolayev]

I wonder if anyone has tried to useML.NET? Still, unlike ONNX, there is already support for .NET in MT5. I wonder if it works when running an EA on VPS.