Machine learning in trading: theory, models, practice and algo-trading - page 3488
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
Just statistics, lots of test samples. But even this is not a 100% guarantee, but still.
But there is no such a lot of test samples - this is often the reason for bad models - lack of data for training.
Except there aren't that many test samples - that's often the reason for bad models - not enough data to train.
Too costly, there has to be another way.
A lot of data for training is not always a good thing either. We get the hospital average.)
Wrong.
MO is looking for patterns, the number of which is limited, about 100, and which can be found with a sample length of no more than 2000. The classification error does NOT go down as the sample length increases. The number of patterns found is a characteristic of a particular set of predictors.
Wrong.
MO looks for patterns that are limited in number, about 100, and that can be found with a sample length of 2000 or less. The classification error does NOT decrease as the sample length increases. The number of patterns found is a characteristic of a particular set of predictors.
CV
Ideally yes, but on market data to get similar models at different parts of the learning curve - I don't recall who has managed to do so.
had some interesting thoughts on MO. exactly how they won't work, didn't write it down in time and forgot about it, big personal information load.
Watching the spinner spin?
I think I'm more level, not interested yet.
I think I'm more level headed, not interested yet.
What's more?