If I have two financial time series and I want to calculate the correlation between them when series1 gives me a negative returns, would that be as simple as picking only those periods where series1 returns are negative and calculating the Pearsons correlation between them? Sorry if its a very basic question as I come from non-stat background.

  • $\begingroup$ What makes the correlation between unfiltered series unsatisfactory for your use case? You can compare both data sets, provided there are sufficient samples, and it's quite likely they correlation ratios will be quite close. $\endgroup$ – Sergei Rodionov Mar 22 at 13:36
  • $\begingroup$ correlation changes from 70% to 50% if I specifically take only days when series1 returns are negative. but that brings to my concern if the interpretation is correct that when series1 is returning -ve the correlation between series1 and series2 is 50%. $\endgroup$ – TRex Mar 23 at 16:36
  • $\begingroup$ I do not think it makes sense to use "correlation" in such a way (in quotation marks as it is no longer correlation by definition). Consider copulas instead. $\endgroup$ – Richard Hardy Mar 27 at 15:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.