Machine learning in trading: theory, models, practice and algo-trading - page 2984

New comment
 
mytarmailS #:

there's an interesting package at the end, you can try to wiggle it.

I'm afraid to get stuck in it and difficulties with conversion to bots will be. I do something similar myself, a complete cycle from pressing the "dough" button to get a bot at the output. Also on the machine, in the region of 10 minutes.

I don't have a team of coders to bolt large foreign bibles to my small tasks.

Maybe onnx version will be made for metac
 
Maxim Dmitrievsky #:

I'm afraid to get stuck in this and difficulties with conversion to bots will be. I do something similar myself, a complete cycle from pressing the "dough" button to getting a bot at the output. Also on the machine, in the region of 10 minutes.

I don't have a team of coders to bolt large foreign bibles to my small tasks.

Maybe I'll make an onnx version for metac
What's onnx got to do with this... Onnx, why are you sticking it everywhere...

Take a ready-made solution and check if it works at all....

If it doesn't, then there is no judgement.....

If yes, then think about your bot and how to place it, if the bot is analogue to bibla....
 
mytarmailS #:
What's this-- Onnx, why do you keep putting him everywhere?

You take a ready-made solution and check if it works at all first...

If it doesn't, then there's no judgement....

If yes, then think about your bot and how to host it

doesn't work

 
Maxim Dmitrievsky #:

It's not working.

Did you try the bibla?
 
mytarmailS #:
Have you tried the bibla?

It won't work. It's for other VRs.

I found something that works, or rather, I invented it. I'm making different variants, seeing what works best.
 

I tried to search through AI what are the variants of local algorithms like KNN and LWLR. He said that there is no such concept at all, and these two belong to the memory-based type, where the training sample is simply stored in memory. Besides these two, he also called memory-based collaborative filtering, but it seems to be the same KNN.

Actually, I wanted to look for a local version of decision trees, but AI directly said that there is no such thing.

I wonder if it makes sense to try to stuff these memory-based ones into an ONNX file, or is it better to do the calculation with MQL tools?

 
Aleksey Nikolayev #:

Tried searching through AI, what are the variants of local algorithms like KNN and LWLR

What do you mean by local algorithm?
What is the problem?
 
mytarmailS #:
What does it mean to say that an algorithm is local?

In the sense in which KNN and LWLR are local. The output depends only on close points, not the whole trayne.

mytarmailS #:
What is the problem?

It doesn't matter, but let it be regression, it's usually easier to deal with.

 
Aleksey Nikolayev #:

In the sense in which KNN and LWLR are local . The output depends only on close points, not on the whole train.

Well, then you can add individual decision trees to your candidate list.
Which are extracted from the regrnsioo nal tree model.

By the way, it's the easiest to integrate into mt.



There could be collaborative filtering, but I'm not sure.
 
Maxim Dmitrievsky #:

It won't work, it's for other BPs.

I'd still try it before jumping to conclusions.
1...297729782979298029812982298329842985298629872988298929902991...3741
New comment