Forecasting rate of return after extreme values when using AR-t-GARCH and QAR-Beta-t-EGARCH

We compare the predictive performances of AR-t-GARCH and recent QAR-Beta-t-EGARCH models. We compare predictive performances for those days when an extreme value is observed, and also for the trading day after each day when an extreme value is observed. We use a historical dataset from the adjusted Dow Jones Industrial Average (DJIA) index. We assume that the forecast users of this study are DJIA options investors. We find that AR-t-GARCH dominates QAR-Beta-t-EGARCH on each day when an extreme value is observed, and QAR-Beta-t-EGARCH dominates AR-t-GARCH on the trading day after each day when an extreme value is observed.