Machine learning in trading: theory, models, practice and algo-trading - page 898
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
Maxim Dmitrievsky, how do you solve this problem?
In general, what are the options? Pieces of the apple can be different...
For each ns from the ensemble add one or another context and use these contexts in some control ns?
By context I mean, for example, a link with some basic definition, concept, predictor and plus some data...
I do not understand the essence of the problem :)
I do not understand the problem :)
What Alexey voiced, probably, can be solved only by optimization...
I added one more predictor - regression - and again the model saw the sense in it - whatever you feed it, it keeps growing...
Buy target
Sell target - things are worse here - a half of entries is missing, but there are not many erroneous ones.
Pay attention to the almost symmetry of the targets - it looks strange.I found another method of correlation in the program, such as the matrix method - maybe it is more correct than the usual?
I think I found it...
Buying
Selling
We can go to bed now.
I got a little crazy and made a forest of scaffolding, now you can do dropouts of individual scaffolding :)
The whole model can now take up to 300 mb on disk, after training, and you can up to a gigabyte
And what's the use of such a model on an untrained history?
I added one more predictor - regression, and again the model saw the sense in it - whatever you feed it, it keeps growing...
So it does not care what to overfit :) It needs new data
So it doesn't care what to overfit to :) it needs new data
I don't know how to load new data for the model in the program, that's why I have to transfer it to MT5, but it's a hassle... I mean I do it, but once a day - to see how it pours.
If i'm wrong i may have some error in predictor, i am recalculating it now.
everything is perfect on the history and with simpler ones, on new data problems arise ) I'll drop it hard and see what happens
I think that all this is an invention of new scaling systems that work in a stationary system and +/- outside of it, and then it's over...
I think that all this is an invention of new scaling systems that work in a stationary system and +/- outside of it, and then it's over...
About the scaling it's for Bat exponent and a direct relative of Erlang himself :))