Machine learning in trading: theory, models, practice and algo-trading - page 2455

New comment
 
Alchemy
 
Evgeny Dyuka #:
The closer the average answer to 1 the better, this is from practice. Why is it so I can not explain, but for me it is one of the main signs of a good model.

Ok, thank you for sharing your experience.

 
mytarmailS #:
Alchemy

alchemy is to race R (or whatever some people who don't understand it) without understanding how its individual packages work and without seeing what's inside them.

I did multicriteria optimization back when you were not on this forum for 6 years, using my own source code and algorithms.

 
Andrey Dik #:

Alchemy is to race R (or whatever some people who do not understand it) without understanding how its individual packages work and without seeing what is inside them.

I was doing a multi-criteria optimization when you weren't on this forum for 6 years, using my own source code and algorithms.

Congratulations, what are the results for 6 years?
 
mytarmailS #:
Congratulations, what results in 6 years?

All sorts, including good ones.

 
Andrey Dik #:

Alchemy is to race R (or whatever some people who do not understand it) without understanding how its individual packages work and without seeing what is inside them.

I did multicriteria optimization back when you were not on this forum for 6 years, using my own source code and algorithms.

Definitely, the code should have been written by your own hand, from beginning to end, without the use of third-party libraries.
+
 
Evgeny Dyuka #:
The closer the average answer to 1 the better, this is from practice. Why so I can not explain, but for me it is one of the main signs of a good model.
The answer is not correct in principle. In my case the sum of all weights of the network coefficients is equal to 1 modulo and if some parameter pulls on itself the greater value of coefficient this coefficient is subtracted from another parameter. That is the sum of coefficients of the network is distributed completely within one. This is one of the ways to reduce overtraining.
 
Renat Akhtyamov #:
Unambiguously, the code should have been written by your own hand, from beginning to end, without the use of third-party libraries.
+

Definitely!!!

You need to write your own neural network, then write your own metatrader, write your own forum, and then it's time to die... genius! when to trade?

 
mytarmailS #:

Definitely!!!

You write your neuronet, then write your metatrader, write your forum, and then it's time to die... genius! When to trade?

you stated a lot of unnecessary things, it's not about that

You didn't even understand what it was about.

 
Renat Akhtyamov #:

a lot of unnecessary information, that's not what this is about.

You didn't even understand what was being said.

I understood everything, I was thinking much further than you, so you didn't understand anything.

1...244824492450245124522453245424552456245724582459246024612462...3695
New comment