Machine learning in trading: theory, models, practice and algo-trading - page 1807
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
If we look at different TFs, the dimensionality becomes more)))).
After preprocessing there are always 5, the rest are highly correlated. If more than 5 it is almost always retraining, if all signs are from the same series
I do not see the point in running deep NS on a pile of correlated traitsAfter preprocessing there are always 5, the rest are highly correlated. If more than 5, it is almost always overtraining, if all the features are from the same row.
I don't see the point in racing deep NS on a pile of correlated traitsThen yes, there is no sense in it if no others are found.
Then yes, there is no point if no others are found.
If time is not taken into account, you can also add seasonality as a feature.
In the form of increments of the period. Somewhere I read that sometimes it turns out a significant feature.
Well that means only 5 rows of signs) Then yes, there is no point if others are not found.
on ZZ the results are nothing
the variant with timeframes is betteron ZZ the results are nothing
variant with timeframes is betterFrom ZZ only the learning zones, not the data.
from ZZ only the training areas, not the data.
yes
yes
I do not understand, after training with more than one TF on the breaks, on the test row breaks are not found?
I do not understand, after training with more than one TF on the breaks, on the test row breaks are not found?
they are not
are not found
Mirage means)))) Or something is missing.
Watch which TF is the best in different combinations remains and go back to the breakdowns and returns.))))
Unfortunately, I was unable to forecast by balance (ZZ built by balance), the results are worse than when forecasting the price itself, there is even nothing to show, I think we should move to the filter of the trades themselves.
By the way, thanks for the perfectly prepared dataset, it is nice to get it working the first time.
I'm not going to tell the truth, I'm going to try to prove that I'm not a bad guy when it comes to studying trading, on the contrary, he is very friendly and "R-people" would be glad to meet a new comrade.
Thanks for the effort.
What ZZ did you use? Can you send me a couple of marked ZZs with different ranges?
I want to group splits now, it will be less combinations.