
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
Hi yes only
By the way, in this version you have programmed everything into a matrix and hence I can't make any changes to the code or easily experiment with anything of my own :))))))
If you can suggest something that needs to be changed in the code to improve the performance, then I can try it myself, because I have a lot of mathematical formulas that can be implemented and tested instead of just using the closing price.
By the way, in this version you have programmed everything into a matrix, and therefore I can't make any changes to the code or easily experiment with anything of my own :))))))
If you can suggest something that needs to be changed in the code to improve the performance, then I can try it myself, because I have a lot of mathematical formulas to implement and test instead of just using the closing price.
What I can suggest is changing things in the code myself. Usually these are non-trivial things based on stream of consciousness and deep learning of machine learning theory. Unfortunately, it is much faster to do it yourself than to explain it, starting from the basics of programming. I can already see that this level is almost out of reach for many people, although this is just the beginning (it's not about bragging, it's about observing people's reactions). So just follow the articles... and study matrices :)
I was surprised how parallel our ideas and developments are! Very useful article!
I understand correctly that to work with your own predictors it is enough to make a modification to this code:
I was surprised how parallel our ideas and developments are! Very useful article!
I understand correctly that to work with your own predictors it is enough to make a modification to this code:
yes, for each individual predictor one element of the array
Hello, where can I get this file?
Hello, where can I get this file?
hello, https://www.mql5.com/en/code/16006
this is the original library from fxsaber, forgot to mention. There is a link in the previous article.
You select predictors one by one and then take the 5 best ones.
The forum thread "Machine Learning in Trading: Theory and Practice (Trading and Beyond)" started with an example that predictors can be decisive in interrelation with each other. True, the example was artificial.
But in Vladimir Perervenko's articles there were examples with identification of interacting predictors on real data. Unfortunately, their detection is very computationally expensive.
You select predictors one by one and then take the 5 best ones.
The forum thread "Machine Learning in Trading: Theory and Practice (Trading and Beyond)" started with an example that predictors can be decisive in interrelation with each other. However, the example was artificial.
But in Vladimir Perervenko's articles there were examples with identification of interacting predictors on real data. Unfortunately, their detection is very computationally expensive.
Is there a method of identifying groups of predictors other than a complete enumeration?
You select predictors one by one and then take the 5 best ones.
The forum thread "Machine Learning in Trading: Theory and Practice (Trading and Beyond)" started with an example that predictors can be decisive in interrelation with each other. However, the example was artificial.
But in Vladimir Perervenko's articles there were examples with identification of interacting predictors on real data. Unfortunately, their detection is very computationally expensive.
at any recombination of features, their importance rarely changes (almost does not change). In the case of constructing new ones from the original ones, yes, it can be computationally complex. Here is a simple example, but it is quite efficient and really selects good features
Usually it is even more trivial - the higher the variance the higher the importance, all features are ranked by variance. That's why this example works well.Is there a method of identifying predictor groups other than a complete enumeration?
Look at Vladimir's earlier articles (2nd or 3rd) where in one of the R packages it was determined using scaffolding. It took a very long time to calculate (many times longer than the training of the main NS), whether it was a full search or some genetics - you should look in the package documentation.
Most likely it was optimised in some way.