Machine learning in trading: theory, models, practice and algo-trading - page 2417

 
Renat Akhtyamov:

even though I gave them the answer.

It's time for a change of direction.

The thread has grown a lot and is full of theory.

I think that the Ministry of Defense will also come in handy and turn golden hands

)))) I agree.

 
Maxim Dmitrievsky:
I have a microsoft autocomplete in my MO-based vskod, and it seems to work. I have not noticed anything alien delightful.

Do not listen and do not read Maxim. It is just a set of clever phrases and there is nothing else he can do. His EA makes only losses and he refuses to comment, but he gives advice to everyone here. Maxim, do something sensible first, like your EA makes a profit, and then give advice.

 
YURY_PROFIT:

Don't listen to or read Maxim. It's just a set of clever phrases and he can't do anything else sensible. His EA makes only losses and he refuses to comment, but he gives advice to everyone here. Maxim, do something sensible first, for example, to make your EA make a profit, and then give advice.

This is a discussion of technical issues, you are going the wrong way!) The comments have long been given in the comments, you can not discuss here Market

 
Maxim Dmitrievsky:

Technical issues are discussed here, you're going the wrong way at all) The comments have long been given in the comments, here you can not discuss the market

Maxim rules.

 
elibrarius:
Similar conclusions - only trees. It's especially fun when maxima are updated and everything shifts. You can of course set maxima manually or automatically (for each feature), but it's crutches.

This alternative is proposed, kinda like normalization is not needed. Not tried yet.

It should also be noted that table data and timeseries are not exactly the same. That's why there might be a hitch on the timeseries.

https://towardsdatascience.com/tabnet-e1b979907694

TabNet: The End of Gradient Boosting?
TabNet: The End of Gradient Boosting?
  • Adam Shafi
  • towardsdatascience.com
Each Step is a block of components. The number of Steps is a hyperparameter option when training the model. Increasing this will increase the learning capacity of the model, but will also increase training time, memory usage and the chance of overfitting. Each Step gets its own vote in the final classification and these votes are equally...
 
Maxim Dmitrievsky:

This alternative is proposed, it seems that normalization is not necessary. I have not tried it yet.

Another thing to note is that tabular data and timeseries are not exactly the same thing. That's why there may be a glitch on the timeseries.

https://towardsdatascience.com/tabnet-e1b979907694

Have you tried it?
 
Maxim Dmitrievsky:

Technical issues are discussed here, you're going the wrong way at all) The comments have long been given in the comments, here you can not discuss the market

You obviously did not understand the point of the post.

Let me explain. You are here talking about smart topics, showing yourself as an incredible professional, while in the comments on your product, dozens of people are asking you: "Maxim, what to do, your robot is leaking constantly, without stopping, can you turn it off, will you upgrade it? And instead of answering these comments and making your product profitable, you are speculating here about technical issues.

And you wrote your long-awaited reply in the comments of your product only after I left a post here.

I want to convey to your interlocutors that your posts are nothing but twaddle, your product is steadily leaking.

And you can not be any moral or ethical right to pretend to be a professional while your product brings losses to hundreds of people.

The tens of articles written by you are an empty sound if at the same time people who have bought your product receive a steady loss from its use.

 
Alexander Ivanov:

Maxim rules.

It doesn't steer at all, only losses.

 
YURY_PROFIT:

It doesn't work at all, only losses.

Yuri, don't believe in fairy tales... all I can say is...
 
YURY_PROFIT:

You apparently did not understand the point of the post.

Let me explain. You are here talking about smart topics, showing yourself as an incredible professional, while in the comments on your product, dozens of people are turning to you with the question: "Maxim, what to do, your robot is leaking continuously, without stopping, can you turn it off, will you upgrade it? And instead of answering these comments and making your product profitable, you are speculating here about technical issues.

And you wrote your long-awaited reply in the comments of your product only after I left a post here.

I want to convey to your interlocutors that your posts are nothing but twaddle, your product is steadily leaking.

And you can not be any moral or ethical right to pretend to be a professional while your product brings losses to hundreds of people.

Dozens of articles written by you are nothing if at the same time the people who bought your product sustainably make losses from its use.

It seems to people who are far from the topic that there is something super mega professional here. In fact, the usual stuff from the field of machine learning, there is nothing to boast about.
Reason: