The contribution of jump signs and activity to forecasting stock price volatility

We propose a novel approach to decompose realized jump measures by type of activity (finite/infinite) and sign, also provide noise-robust versions the ABD test (Andersen et al., 2007b) semivariance measures. find that infinite (finite) jumps improve forecasts at shorter (longer) horizons; but contribution signed is limited. As expected, deliver substantial forecast improvements higher sampling frequencies, although standard volatility 300-s frequency generate smallest MSPEs. Since no single model dominates across forecasting horizon, we show averaged – using time-varying weights models from confidence set generally outperform both benchmark best extended HAR model. Finally, based on transaction are inferior clock-based sampling.