Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Performance difference between various look back periods in relative strength model - How do we know what will perform better in the future?

Hi all,

New to the forum, and to this type of investing. I have been working on a relative strength rotation model, but I have a hard time understanding why the backtests show so different performance results depending on what look back period i put in.

3 month delivers more than twice the return as a 12 month look back period, and that is concerning to me. I would expect more similar returns, since momentum has been shown to deliver results on a 3-12 month look back period.

3 month look back period has produced great results for the last couple of decades, but how do we know it will continue?

Most of us only have very few decades to invest in, and it would be very problematic for ones performance if the optimal look back period in the future differs from what was chosen for the model.

How do you guys handle this in your modelling?

Best regards
Lars

1 response

Just a suggestion, pick several different look back periods (e.g. 3, 6, 9, 12 month), sum the returns, then take the average.
You could even weight the returns differently before summing perhaps giving more weight to the more recent returns.