Machine learning in trading: theory, models, practice and algo-trading - page 121
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
What if a couple of trades turn out to be the main profit, and thousands of others turn out to be small losses? - It's kind of a bad idea.
Exactly right, the drawback of all target criteria for TS - they do not take into account the number of deals. That is, on a small number of trades you can randomly obtain a large value of the criterion.
Well, why all of them... You can add arguments to the formula that take into account the ratio of profitable/loss-making trades and the total number of trades.
I've been through the whole thread.
You guys are digging in all the wrong places.
It turns out that there is a tool to evaluate R models
How to evaluate R models in Azure Machine Learning Studio
I've been through the whole thread.
You guys are digging in all the wrong places.
And where to dig?
I join the question.
Yes Vadim where to dig?
p.s. glad you were unbanned
I've been through the whole thread.
You guys are digging in all the wrong places.
Welcome out of the ban, first of all. I hope I don't have to complain about you again.
You are our demokrat.
Where do I dig?
He created a separate thread for diggers. See Machine Learning: Theory and Practice (trading only; entry only for those who have a state)
It turns out that there is a tool for evaluating my models
How to evaluate R models in Azure Machine Learning Studio
I understood that it's just a nice visual programming shell, inside all those "modules" is the usual R code. Basically, the data is divided into two groups train/test, and the model is trained and tested as standard, all the same can be done in rattle.
More subscription to the service is expensive, 100sd for 25 hours of calculations per month. Will save only if their cloud is more productive, and an hour does much more than the usual computer.
At the end (Evaluate module) for the model is calculated <<"Accuracy", "Precision", "Recall", and "AUC">> etc., there is nothing new here too.
I understood that this is just a nice visual shell for programming, inside all those "modules" is the usual R code. In essence, the data are divided into two groups training/test, and the model is trained and tested as standard, all the same can be done in rattle.
More subscription to the service is expensive, 100sd for 25 hours of calculations per month. It will help if their cloud is more productive and they do much more calculations per hour than a usual computer.