A semi-markovian approach to drawdown-based measures

In this paper we assess the suitability of weighted-indexed semi-Markov chains (WISMC) to study risk measures as applied to high-frequency financial data. The considered measures are the drawdown of fixed level, the time to crash, the speed of crash, the recovery time and the speed of recovery; they provide valuable information in portfolio management and in the selection of investments. The results obtained by implementing the WISMC model are compared with those based on the real data and also with those achieved by GARCH and EGARCH models. Globally, the WISMC model performs much better than the other econometric models for all the considered measures unless in the cases when the percentage of censored units is more than 30% where the models behave similarly.