Machine learning in trading: theory, models, practice and algo-trading - page 2641

 
Maxim Dmitrievsky #:
Not comparing, just having fun 😀
Well python is the same as R, would have crossed over somehow, Vladimir Perervenko crossed over kinda, at least attempts were
Articles topic yes, especially on MO. There is very little normal material for trading.

Yes, I would switch, but people switch because of some advantage, and I don't see it. In terms of data processing, python is empty compared to rca.

Rca has everything that python has (and much more), but python doesn't have what Rca has).

The advantages of python is that it is a general-purpose language, you can make ehe-sheets, it's cool, more readable code and growing community, but so on the rest of the dozens of parameters python is inferior ....

All the coolness of Pandas with its dataframes, vectorisation, etc... is a stolen standard Rki syntax ...

Okay, I'm getting carried away, I'm a matter of taste...

 
mytarmailS #:

I would switch, but people switch because of some advantage and I don't see it. In terms of data processing, python is empty compared to rca.

Rca has everything that python has (and much more), but python doesn't have what Rca has).

The advantages of python is that it is a general-purpose language, you can make echoes, it's cool, more readable code and a growing community, but so on the rest of the dozens of parameters python is inferior ....

All the coolness of Pandas with its dataframes, vectorisation, etc... is a stolen standard Rky syntax...

Okay, I'm getting carried away, I.P. is a matter of taste.

Ramp is like a separate language with its own tricks.
 
Maxim Dmitrievsky #:
The ramp is like a separate language with its own jokes.

and everybody loves it.)

 
Maxim Dmitrievsky #:

In the end, Max, everything can be represented in the form of rules and put into any language.

The main thing is to have something to pour in, so you can and should develop TC on the language you are comfortable with


And python, yes, I need to learn it... the future is behind it, but I am so lazy... and it is awfully raw in terms of the tasks that Rka solves.

I think the most important thing in trading is the speed of implementation and testing of your ideas.

 
mytarmailS #:

Ultimately Max, everything can be represented as rules and populated into any NPS.

The main thing is to have something to pour in, so you can and should develop TC in the language you are comfortable with


And python, yes, you should learn it... the future is behind it, but you are so lazy... and it is awfully raw in terms of the tasks that RCA solves.

I think the most important thing in trading is the speed of realisation and testing your ideas.

I have spent some time thinking up new signs based on increments (I wrote a message before, but nobody suggested anything). Maybe I'll show the results soon. My approach is a bit different - I'm going through the tags and trying to find informative features for them. That is, I'm looking for good feature-tag pairs. It's basically the same thing, but viewed from the side. I rely on rules, but on stationarity and stability of signs. Or rather, I'm looking for a compromise between stationarity and informativeness.

I use only catbust and enthusiasm, you can do it on any jap)
 
Maxim Dmitrievsky #:
I've been spending some time coming up with new features based on increments (wrote a post before this one, but nobody suggested anything). Maybe I'll show you the results soon. My approach is a bit different - I'm going through the tags and trying to find informative features for them. That is, I'm looking for good feature-tag pairs. It's basically the same thing, but viewed from the side. I rely on rules, but on stationarity and stability of signs. Or rather, I'm looking for a compromise between stationarity and informativeness.

Yeah, yeah, something similar, but you can go further with the rules, as I see it.

 
mytarmailS #:

Yeah, yeah, something similar, but you can take it further, I think.

Boost has a new feature, you can drop bad signs. I haven't had a chance to use it yet.

https://catboost.ai/en/docs/concepts/python-reference_catboost_select_features

select_features
select_features
  • catboost.ai
Purpose. Select the best features and drop harmful features from the dataset. Method call format.
 
Maxim Dmitrievsky #:

Boost has a new feature, you can drop bad signs. Haven't had a chance to use it yet

https://catboost.ai/en/docs/concepts/python-reference_catboost_select_features

I doubt it's gonna do anything.

 
mytarmailS #:

I doubt it's gonna do anything.

Combined with other methods, it might do something.
 
Maxim Dmitrievsky #:
Combined with other methods, it might do something.
How's that?