Machine learning in trading: theory, models, practice and algo-trading - page 389
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
you don't need to train machines, you first need to have iron nerves and connections in the higher echelons of power to make a profitable trade
Maybe, but this is the way for tough, brutal men, and there should be a way for humble intellectuals...
Forum on trading, automated trading systems and testing of trading strategies
Computer learning: theory and practice (trading and not only)
SanSanych Fomenko, 2017.06.09 15:17
No, of course not.
I'm busy getting some guarantees for some future.
Hi there!!!! I'm glad this thread hasn't died out and still is alive and well, so I have a question for audience. I have dataset for learning but unfortunately it became so big that learning takes too long. Can someone build a model with his work, and then we will see how it works together!!!!!.
You decided as a numer.ai to throw sets that you would build a model? :) Use azure machine learning studio for verification, it works fast in the cloud.
If I have time, I'll build a model later, I'll show you how to do it. In principle, there is a detailed help and examples, you can easily do everything without programming
Then you mount your trained model in the cloud and fetch the results with a bot via webrequests
https://azure.microsoft.com/ru-ru/services/machine-learning/
You decided as numer.ai to throw sets that would build your model? :) Use azure machine learning studio for verification, it quickly counts in the cloud.
If I have time, I'll build a model later, I'll show you how to do it. In principle, there is a detailed help and examples, everything can easily be done without programming
Then you mount your trained model in the cloud and fetch the results with a bot via webrequests
https://azure.microsoft.com/ru-ru/services/machine-learning/
Now I will try to figure it out.
I will try to figure it out.
I'm already doing it, I'll report later.) The only thing you'll need to do is to make a set with a "," instead of ";" delimiter for the studio, otherwise it won't identify columns correctly
There is a good manual with examples in Russian
Set originally for classifier Reshetov prepared? By the way, I would like to rewrite it in mql5 for OpenCl.
Can someone build a model with his own developments, and then we will see how it works together!!!!!
blue decision trees, red Boosted decision trees, comparison of 2 models, training/test sampling 75\25
Now just upload the experiment to the studio and play with it https://gallery.cortanaintelligence.com/Experiment/BuySell-created-on-10-06-2017.
You can just change training models, substitute other blocks, for example neuronet, do a features filter, some transformations over features, etc.
For the neural network you will need to do some normalization, because the decision trees work without it.
I have not been able to train the model, the predictors and the target don't add up.
Now just upload the experiment to the studio and play with it https://gallery.cortanaintelligence.com/Experiment/BuySell-created-on-10-06-2017.
You can just change training models, substitute other blocks, for example neuronet, do a features filter, some transformations over features, etc.
For a neural network you will need to do some normalization as well, since decision trees work without it
The experiment is good, but then how to use its results in MT? I can unload the model and put it into an indicator to see how it works in the out-of-sample area????