Machine learning in trading: theory, models, practice and algo-trading - page 3715

 
Maxim Kuznetsov #:

and now you can do that? ;-) some people have had "iron nails" for the other one.

or does it mean some other account, different from some other account that was previously shown in the screenshots?

Trading accounts are different, this is a picture from the terminal, it looks like a signalling one there
 
Maxon, is there an EA without python that works on moe? simple and practical ok
 
Altman, in an interview with The Verge, acknowledged a bubble in the AI market.
 
Aleksey Nikolayev #:
Altman in an interview with The Verge recognised a bubble in the AI market.
Exaggerated expectations of investors are the way to their losses)
 
Aleksey Nikolayev #:
Altman, in an interview with The Verge, acknowledged a bubble in the AI market.
It's funny, pro-mptus people only notice obvious things when they are reported by pro-mptus authorities. It's a sect, though.
[Deleted]  
Ivan Butko #:
maxon, is there an EA without python, working on moe? you know, to make it simple and practical ok

Shalom! Of course there is.

 
APXuTEKTOP #:

Shalom! Of course there is.

Where do you get it?

 

A small portion of realism about AI from one of its creators. The translation is not very good, but the meaning is conveyed. The (extremist) Meta adverts are a bit annoying, but not enough to drown out the main theme.

The basic idea is that the current AI is pretty good at "understanding" language (predicting words) because of its discrete nature. It does worse with reality video because the physics of reality is continuous, not discrete.


 

Imho, it is worth trying to revitalise the thread - it was quite good.

But I want something new, the old topics on MO have already become a bit boring.

Such a new topic could be probabilistic machine learning. This is an approach in which the output of an algorithm is not a specific numerical value, but its probability distribution. Imho, this approach is more organically suited to the task of trading.

I will write about this topic as I am in the mood.

PS. I have nothing against if someone will write in the thread on any other topics related to MO.

 
Aleksey Nikolayev #:
probabilistic machine learning

In the context of familiar MO approaches (classification and regression), probabilistic MO is a generalisation of them.

1) Classification - the output of the algorithm is a discrete distribution of the class number.

2) Regression - the output of the algorithm is a normal distribution with a mean dependent on the inputs (perceived as a prediction) and a fixed variance independent of the inputs (perceived as a prediction error).