Machine learning in trading: theory, models, practice and algo-trading - page 829
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
And where did you find CNTK, if it's not a secret, there is like a DEAP library of evolutionary programming and there is not much trouble and tons of code, although there are no generated examples, but that would be too much?
It's no secret. Open any source code in Python.
There are no neural networks. This is genetic programming.
What about here? Or am I looking at the wrong thing?
For neural networks I still choose CNTK. The reasons are the following. In Python I train the network and store the trained network. In C++ the DLL for MT which uses a trained network. That's because of the convenience of connecting a trained network and I choose. Tensorflow in C++ for Windows is a terrible tambourine dance.
Of course it's the developer's choice. For me so keras is a perfect start. True, I use keras for R, and no dancing. I am finishing an article in which I will show this variant.
Maybe in your blog here you can briefly sketch a block diagram of your implementation of genetic programming?
Good luck
What about here? Or am I looking at the wrong thing?
That was a long time ago. I gave you a link to another repository. This repository is not for this forum. Where I wrote about NS, I wrote that I will answer all questions on the code.
The answer you gave and it is clear. Good luck
This is from game theory, it seems. That's how the trader plays his game, and the market plays its game.
In fact the data from the past is only used when searching for points (tops and troughs), and the returns or classes based on data from the future (to the right of the predicted tops) are considered for the tags themselves.
So, in fact, there's no peeking in the training data.
Nevertheless "coryphaei" continue to sell us this fake under the guise of secret knowledge, but promised proofs can not provide, probably the same thing with metrics ...
The most reliable secret is when you don't know or don't understand, here it's been said a lot that ZZ is the wrong target because it supposedly peeks into the past.
And even more reliable - when you did not know, and then even forgot :)
The most reliable secret is when you don't know or don't understand, it was said here a lot that ZZ is a wrong target because it supposedly peeks into the past.
In fact, the data from the past is only used when searching for points (vertices and troughs), and the returns or classes are considered for the tags themselves, based on data from the future (to the right of the predicted vertices).
That is, in fact, there is no peeking in the training data.
I'm sorry for grumbling about it for nothing... Nevertheless "coryphaei" continue to sell us this fake under the guise of their secret knowledge, but they can`t provide us promised proofs, maybe the same thing with metrics...
I stand corrected, wise guys did it, I admit my mistake, ZZ is cool targeting, RSI is cool features, I apologize for grumbling for nothing.
I stand corrected, wise ones pointed it out, I admit my mistake, ZZ is cool targeting, RSI is cool features, I apologize for grumbling for nothing.
rsi is not much different from returnee )
I stand corrected, wise ones pointed it out, I admit my mistake, ZZ is cool targeting, RSI is cool features, I apologize for grumbling for nothing.
It's not grumbling, it's idle chatter.
Two days waiting for the results of the run, which confirm your words.
PS.
And do not write off as last time