Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Uncorrelate those models - Seattle/Portland meetup talk

Something that I found myself running into when modeling is that sometimes my explanatory variables were explaining the same thing, and appeared to be correlated over time. I figured that I wasn't the only person running into this problem, aware of it or not, and decided to do a little research into it. I'll post a notebook shortly with the code for some of the methods I used. Hopefully these backtests demonstrate the power of adding uncorrelated factors to your models.

Here is the code for the backtests that I talked about during the Seattle and Portland meetups. The final model does require access to the paid Sentdex dataset though the theory should still hold with the free data. Note that these algos are for demonstration/education purpose so commission and slippage have been disabled.

2 Correlated Factors

Quality, Value

Disclaimer

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.

5 responses

2 Uncorrelated Factors

Value, Momentum

3 Uncorrelated Factors + new datasource

Value, Momentum, Sentiment

As mentioned above this algo uses the paid version of the Sentdex data, though you can easily run the algo with the free version. Just change the import statement from

from quantopian.pipeline.data.sentdex import sentiment as sentdex  

to

from quantopian.pipeline.data.sentdex import sentiment_free as sentdex  

also don't forget to adjust the date ranges if using the free dataset.

James,
Nice talk in Portland...thanks!
Looking forward to your notebook posting. I found that very interesting.
alan

Looking forward to learning from this. Thanks for sharing.

Hey All - really dropped the ball on this one! Sorry for the delay I spent most of the summer heads down on Alphalens. Here is the NB for the Seattle and Portland talks it is pretty slow to execute (lots of permutations and correlations) so it may be better to just view the preview, otherwise enjoy.

And I'll be in Seattle with Max on October 10th (same venue) for another talk!