Machine learning in trading: theory, models, practice and algo-trading - page 1154

 
The grail:

The "fractality" is overcome by vector forecasting, i.e. when the increments are vangued at a time, for example for 10 minutes forward, an hour, 6 hours, a day, a week

Have you tried it? Have the results improved?

Grail:

"head and shoulders" is a price pattern, like every algotrader wrote a pettern correlator to make sure that the market does not see patterns, there is no statistical justification that some "figure" can predict something, they are "visible" just like "animals" in clouds, this is pareidolia

The "head and shoulders" was an aligory, meaning patterns that we don't know about but search for with networks, etc. And if you imagine that they exist, it is obvious that they do not exactly repeat themselves, they will be different in time-chat-amplitude and therefore the network will not find them, and not because it is stupid, stupid one who shoves raw data into it and expects a miracle

 
mytarmailS:

Have you tried it that way? Have the results improved?

The "head and shoulders" was an aligory, meaning patterns that we don't know about but are looking for with networks, etc. And if you imagine that they exist, it's obvious that they don't exactly repeat themselves, they will be different in time-chat-amplitude and therefore the network will not find them, and not because it's dumb, dumb is the one who put the raw data into it and expects a miracle

he who knows how to make the data "not raw" in networks does not need...

 
Maxim Kuznetsov:

Someone who knows how to make data "not raw" in networks does not need...

agree

 
Grail:

ZZ as a Target - facepalm


Authoritative, unapologetic and most importantly what definitions...

 
Vladimir Perervenko:

Authoritative, unapologetic and most importantly what definitions...

Oh Vladimir has appeared! Why you ignore me Vladimir? It's not nice of you to do that

 
mytarmailS:

Oh Vladimir has appeared! Why do you ignore me Vladimir? It's not nice of you

No, I'm 3 months far abroad (in the jungle) and not always have access to the Internet. Today is a happy occasion. I will be back after the new year, I will continue your question.

Good luck

 
Vladimir Perervenko:

No, I'm three months away from home abroad (in the jungle) and don't always have access to the Internet. Today is a happy occasion. I'll be back after the new year, I'll continue your question.

Good luck

Ok, good luck to you too.

 
mytarmailS:

If p-studio.


I'm doing all this, but I get an error

> install.packages("F:/FX/R/catboost-R-Windows-0.7.2.tgz", repos = NULL, type = .Platform$pkgType)
Warning in install.packages :
  error 1 in extracting from zip file
Warning in install.packages :
  cannot open compressed file 'catboost-R-Windows-0.7.2.tgz/DESCRIPTION', probable reason 'No such file or directory'
Error in install.packages : cannot open the connection

The file is unzipped by the archiver without problems - I checked.

What's wrong with it?

 
mytarmailS:

Have you tried it that way? Have the results improved?

The "head and shoulders" was an aligory, meaning patterns that we don't know about but are looking for with networks, etc. And if you imagine that they exist, it is obvious that they do not exactly repeat themselves, that is they will be different in time-chat-amplitude, and therefore the network will not find them, and not because it is dumb, dumb is the one who put the raw data into it and expects a miracle.

I've improved on tests, I don't use neural networks and patterns, I work with current state of the whole market, the forest mostly.

 
Aleksey Vyazmikin:

I'm doing all this, but I get an error

The file unzipped with the archiver without problems - I checked.

What's wrong?

Windows or Linux?

Reason: