Covariance forecasting in equity markets

摘要: We compare the performance of popular covariance forecasting models in context a portfolio major European equity indices. find that based on high-frequency data offer clear advantage terms statistical accuracy. They also yield more theoretically consistent predictions from an empirical asset pricing perspective, and, lead to superior out-of-sample performance. Overall, parsimonious Vector Heterogeneous Autoregressive (VHAR) model involves lagged daily, weekly and monthly realised covariances achieves best out competing models. A promising new simple hybrid estimator is developed exploits option-implied information while adjusting for volatility riskpremium. Relative does not change during global financial crisis, or, if different forecast horizon, intraday sampling frequency employed. Finally, our evidence remains robust when we consider alternative sample U.S. stocks.