I have to model the distribution of past returns for several time horizon using a gaussian mixture distribution. I first build my time series of past returns thanks to a rolling sum of my past log returns, then fit a gaussian mixture on it.

How would you decide which history of past returns to use to fit your model ? My first guess was to try several history and then take the one maximizing my log likelihood, but I feel like I overfit my data with this approach


Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.