Retraining

New comment
 
Who uses what techniques to minimise retraining of the model/advisor?
 
Can they (these methods) be, exist? It depends on the strategy, on the EA itself. Either it only adjusts itself in the tester and drains, or it optimises itself and works.
 
Aliaksandr Hryshyn:
Who uses what techniques to minimise retraining of the model/advisor?
If you understand how it works, you optimise the main 1-3 parameters. Then you optimize the secondary ones.
 
Dmitry Fedoseev:
Can they (these methods) exist? It depends on the strategy, on the EA itself. Either it only adjusts itself in the tester and fails, or it optimizes itself and works.
Yes, there can be. For example, setting limits on model complexity. Of course, the MT4/5 strategy tester may lack features. Or one may try to compare results distributions on training and test sets. Maybe someone has already conducted research in this direction.
 
A similar thread, only without the blah-blah
 
СанСаныч Фоменко:
A similar thread, but without the blah-blah-blah.
This is exactly where there is only one blah-blah-blah, it is in that branch, solid bababolas, at least one thing is good - philosophical conversations "for stationarity" are over. SanSanych, maybe it is time for you to stop your subtle rudeness? Here in this thread only two people checked in and only one post written, and you're about blah-blah write.
 
Aliaksandr Hryshyn:
Yes, there can be. For example, setting limits on model complexity. Of course, the MT4/5 strategy tester may lack features. Or one may try to compare results distributions on training and test sets. Maybe someone has already conducted research in this direction.

Who will install it and how?

The training set and the text set, that's another topic.

 
If a forward test is used, "overtraining" will be visible by degradation of results outside the optimization period. Unfortunately, MT does not provide any built-in methods of comparing optimisation and forward test results, i.e. it is suggested to do it manually (by eye), by external programs or scripts at one's own discretion.
 
Dmitry Fedoseev:

Who will install it and how?

Training and text sets, that's another topic.

Why is it different? Because it is broken down to assess overtraining or it is part of the method itself.
 
Dmitry Fedoseev:
The only place where there is only one blah-blah-blah is in that branch, all of them are blabbermouths, at least one thing is good - philosophical conversations "for stationarity" are over. SanSanych, maybe it is time for you to stop your subtle rudeness? Here in this thread only two people checked out and only one post each, and you're blah-blah write.
Blah-blah refers to content, not people. If you took it as referring to people, I apologise.
 
СанСаныч Фоменко:
Blah blah refers to content, not people. If you took it as referring to people, I apologise.
And I thought "blah-blah" was judged by what they write, not by the people.
1234567
New comment