I am simulating through Monte Carlo, multivariate correlated returns of different products composing an Oil&Gas portfolio. The historical prices (from which I computed the log-returns) of the various products present different time horizon (most of them are daily, but some are monthly). For sake of coherence how could this be solved? Considering that my aim would be keep working with daily data.

  • 2
    $\begingroup$ My guess is that you indeed will need to work with the lowest frequency. Meaning that if there is a variable that you want to include in your dataset and that variable has monthly frequency then you need to work at monthly frequency. That is for instance one of the issues with macroeconomic data which only exists at quarterly frequency and therefore most models are calibrated quarterly. There are ways of converting processes (such as ARs) to different frequencies, but not returns. $\endgroup$ – phdstudent Jun 7 '18 at 12:51
  • $\begingroup$ For maximising the informational content of your data could you not evaluate the correlations of daily series daily but for lower frequency data resort to that frequency for assessing the correlation. To convert the variance of the series to daily can you not just divide the variance of the monthly series by 21 trading days? $\endgroup$ – Attack68 Jun 7 '18 at 17:12

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.