Machine learning in trading: theory, models, practice and algo-trading - page 3408

New comment
 
Forester #:
A general reference, ChatGPT style. The best way to show off your encyclopaedic knowledge.
Aha, also referring to Alexey's post from 2016 in 2024 is also something, only Sanych could do such a thing))))
I was sure it was a bot at first, but I looked at the profile, it seems to be a person. Though it could be anything.
 
And who is involved in crypto, you can issue your own token, can't you?
 
Maxim Dmitrievsky #:

I decided to devote the weekend to the study of numeraa, because I could not get my hands on it.

And how
 
mytarmailS #:
And how
So far only uploaded the 1st model from the tutorial, I was credited 0.1 local currency. Didn't have time for the 2nd tutorial :)

There are 4 gig datasets, 2000 features.

I understand, there you stack models, the more you stack the more you will be charged, after checking on new data. All models from managers they stack into one big one and it sort of trades somewhere on the exchange. According to their formulas they estimate the contribution of each model to the total result.
 
Aleksey Vyazmikin #:

Well, graphs speak better than words here - in brief, there is no special effect of improvement (relative to the original), but at the same time,"linear attribute screening" has shown itself to be better, if measured by the average value of the balance of models on an independent sample. At the same time, balancing and abesssifting were able to isolate significant predictors on which we can already build a model - we can consider that the backbone. It may be worthwhile to do a dozen balances in a cycle and pull out all the resulting predictors.

One thing is obvious: these methods are fast, but far from optimal.

Numerai says above in the video that there are no perfect methods.

They call their approach feature neutralisation, looking at correlation between features and labels and std. In short, using Sanych's method

https://colab.research.google.com/github/numerai/example-scripts/blob/master/feature_neutralization.ipynb#scrollTo=meowEBs-PwtB

Here you go, at the same time you can pull up python :) and your datasets are about as big as theirs.
 

Probably good stuff. From practitioners, not academics like Prado.

Although, they also slipped in the Embargo method in the previous tutorial :)

 
Maxim Dmitrievsky #:

Have you seen the results of their fund?

Do they take money for something? It may turn out that the main earnings are not from the market at all.

 
Rorschach #:

Have you seen the results of their fund?

Do they take money for something? It may turn out that the main earnings are not from the market at all.

I haven't figured it out yet. There should be information on the website
 
Rorschach #:

Have you seen the results of their fund?

Do they take money for something? It may turn out that the main earnings are not from the market at all.

https://numerai.fund/

something is shakily traded there... but I am more interested in what can be earned on models.

Yep, they take"To stake NMR on your model you must first deposit NMR into your Numerai wallet at your account's unique deposit address." So you bet on your own model with their crypto, and according to the results they either give you money or take away yours :)
Numerai Hedge Fund
  • numerai.fund
Numerai is a quant hedge fund built on thousands of crowdsourced machine learning models.
 
So far from interesting is about feature neutralisation... further tutorial on ensembles
1...340134023403340434053406340734083409341034113412341334143415...3656
New comment