This a reference request for any python notebooks, packages or blogs that teach how to do asset allocation using multivariate copulas. How can copula portfolio optimization actually be implemented in code rather than in theory?

  • $\begingroup$ I do not understand why one would be interested in doing mean-variance optimisation with copulas. Isn't the whole point of using variance as a risk measure the fact that in a multivariate normal world it is the only risk measure? And the point of copulas is going beyond multivariate normal? $\endgroup$ – g g Aug 24 '20 at 18:56
  • $\begingroup$ distributions don't have to be multivariate normal to have a mean and covariance $\endgroup$ – develarist Aug 24 '20 at 18:59
  • $\begingroup$ Of course but the relevance of covariance is greatly reduced if you do not have multivariate normals. $\endgroup$ – g g Aug 24 '20 at 20:53
  • $\begingroup$ i've removed the word mean-variance from the question $\endgroup$ – develarist Aug 24 '20 at 20:59
  • $\begingroup$ @gg, if the agent is assumed to have quadratic utility, mean and variance are the only two moments that matter in maximization of expected utility -- regardless of the distribution. This is one of the justifications behind mean-variance optimization. $\endgroup$ – Richard Hardy Aug 27 '20 at 11:20

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.