Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Help with creating a feature set for a machine learning algo
Zach D
posted Apr 11, 2015

Hey everyone,

I am coding a simple machine learning algorithm for trading and I could use some help with picking features to use in my code. I am under the impression that I can use chart technical indicators; examples in the "Chart Technical Indicators" section in this link to the MathWorks financial toolbox. I want to use a feature set to predict future price direction, but I'm not sure which features are most relevant and what number or periods they should be calculated for.

I already coded the majority of the algorithm in MATLAB, and the algorithm is reasonably good at making predictions, but I am running into some issues with overfitting, and I suspect this is because my feature set could use some improvement. I have started translating my code to Python because I want to backtest it here, but my feature set just isn't quite good enough yet for me to post my Python code. Any help with what features I should focus on would be greatly appreciated.

4 responses
Sol Marte
Apr 11, 2015

Machine learning is curve-fitting by nature. You need to do a lot of out of sample testing.

James Jack
Apr 11, 2015

Zach I'd be interested to know if you have any positive results once you've coded it in Python. I did a quick run of this a while ago, using a very basic selection of RSI, ATR, ROC, etc, and found it had absolutely zero predictive power using a wide selection of classifiers.

I thought about combining into an event study but never got round to it

Lucas Silva
Apr 11, 2015

Some say TA had some predictive power 15 years ago, not anymore. But this is just anecdotal evidence, most research I read that says TA worked were very poor and didn't count for all the over-fitting factors and were not rigorous enough.

Zach D
Apr 11, 2015

James, I've had fairly positive results in MATLAB trading Bitcoin, but I think this is partly because there are so many inefficiencies in Bitcoin markets so they are easy to predict.

About Quantopian
Careers
Community
Events
Documentation
Academia
Lectures
Status
Facebook
Twitter
LinkedIn
YouTube
Terms of Use
Privacy Policy

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian.

In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.