Machine learning in trading: theory, models, practice and algo-trading - page 1326
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
The sample is stationary, the breakdown for training has changed, but the breakdown for independent evaluation has remained the same.
Please expand on your point.
it's just a sampling for a specific situation and specific conditions
i.e., it's just a sampling of options that fall out at random and will not be replicated in other subsamples
see details in the statistical analysis
it's just a sampling evaluation for a specific situation and specific conditions
i.e., it's just a random sampling of options that will not be replicated in other subsamples
See statistical analysis for details.
This can be said about everything we do here.
What do you suggest we do to make the results more valid?
Maybe take another tool and look at the results on it?
repeat with a different seed to see if it's random or not
Now I decided to flip the samples - lern becomes validation. And then I'll try to change the seed.
That can be said about everything we do here.
What do you suggest we do to make the results more valid?
Maybe take another tool, and look at the results on it?
Test it on independent samples, and if the results are normal, then run the model and go smoke bamboo
make the model selection process as less time consuming as possible and relax
test on independent samples, if the result is normal, then run the model and go smoke bamboo
Make the process of model selection as time-consuming as possible and relax
What does it mean to use independent samples in this context?
What do you mean by independent samples in this context?
Well, on a test sample, which has already been defined, pardon the tautology.
in short, do everything as written in the books
no need to invent any bullshit. it all does not workDamn well on the test sample, which definition has already been determined, excuse the tautology
to do everything as it is written in the books.
there's no need to make up any wild stuff. it all doesn't workAh, I see, someone is not reading carefully :)
Forum on trading, automated trading systems and trading strategies testing
Theories and Practices of Machine Learning in Trading (Trading and not only)
Aleksey Vyazmikin, 2019.02.13 07:27
...
Testing of models was conducted on a fixed sample - 2018, not participating in the training.
...
Ah, I see, someone does not read carefully :)
What should I do? Bruteforce the models -> choose the best -> trade
Or something should happen at the level of existential enlightenment?
What's the brain to do? put a bruteforce on models -> choose the best one -> trade
Or there should be something at the level of existential enlightenment.
Well, I'm in favor of finding the right method for training, the settings there, in order to raise the averages, and then trade by them. I have not tried all variants yet - it's too early. For now the last year herbarium is trading - the indicators caused by the error in predictors have considerably worsened the model but it is still profitable in the test.