The in-sample results show that all nonlinearities are highly signicant and improve
the description of the data. The out-of-sample results show that for shorter
horizons, up to 10 days, accounting for these nonlinearities signicantly improves the
forecast performance compared to a linear ARFI model. Such short-term volatility
forecasts are especially useful for short-term risk management, including Value-at-
Risk. For longer horizons no benet is obtained from incorporating nonlinearities