I've got the data of one trading strategy that generates a long-short portfolio, from year 2000 to 2024, and I've used the portfolio return from 2000 to 2020 to run the Fama-french five factor model to test the alpha (also the regression results). Now I want to do a out-of-sample prediction using data from 2021 to 2024, and I'm confused what to do:
First I know I can get a fitted Rp_hat with the estimated alpha, betas and five factor data from 2021 to 2024;
But what to do next? Should I test if the difference between Rp_hat and real Rp is significant from zero? Should I use this difference to run the Fama-french five factor model again to see if the difference between estimated result and reality can be explained by the risk factors (i.e. test if still exist significant alpha?) Or any other method that is correct?
Thanks for answering,
Carlos
First I know I can get a fitted Rp_hat with the estimated alpha, betas and five factor data from 2021 to 2024;
But what to do next? Should I test if the difference between Rp_hat and real Rp is significant from zero? Should I use this difference to run the Fama-french five factor model again to see if the difference between estimated result and reality can be explained by the risk factors (i.e. test if still exist significant alpha?) Or any other method that is correct?
Thanks for answering,
Carlos
