Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Backtesting - forward vs backward
Saki sakili
posted

Team,
quick question. Doing backtest for one of my strategies to determine if my approach is valid or completely flawed.

I do backtesting for last 1 year and if model is fine then I do backtest for last 7 years.

If model shows it worked well for last 7 years then I consider it valid.

I know this is in contradiction to the theory where you have to first do tests for the past period and then verify with future data sets.
What would be difference in this case?

Any feedback is appreciated.
thanks in advance

4 responses
Mohammad Forouzani

I think the theory is that older data is less relevant than newer data.

So if you test your system on a year of data from 7 years ago, and then get it to run on last years data profitably, then it is better than vice versa.
I.e. do all your overfitting etc on old data and see if it would still work on new data.

Keep in mind that the markets are always evolving, so many algos that wouldn't work many years ago, may work today. I think is part of the reason that recent data is more relevant.

Saki sakili

Thanks for your feedback.
I fully agree with you regarding what might be ideal approach.
In this particular case, if I have situation where previous data works as good as newer set, question is would that be valid approach since it obviously works or I should completely disregard that just because theory says otherwise and look for another way to get to the same conclusion?
I didn't do any overfitting for past periods so it seems like valid approach, right?
Thanks in advance

Mohammad Forouzani

The only way you know if you did overfitting is in hindsight, there is no way for you to know in advance.

Once you are comfortable with your algo, the only way to really know how good it is, is to trade it live (after some paper trading). But only in a live environment can you really see the true cost of slippage, commissions, interest on leverage or shorts etc. These all have to be accounted to see if your algo is actually profitable.

Saki sakili

Mohammad, thanks again for your feedback. I guess, the only way to know is to test it in live environment.

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.