The adjusted measure of realized volatility suggested in  is applied to high- frequency orderbook and transaction data of DAX and BUND futures from EU- REX in order to identify the drivers of intraday volatility. Four components are identified to have predictive power: an auto-regressive pattern, a seasonal pattern, long-term memory and scheduled data releases. These components are analyzed in detail. Some evidence for two additional components, market microstrucuture events and unscheduled news, is given. Depending on the sampling frequency we estimate that between one and two thirds of the variation in realized volatility can be predicted by a simple linear model based on the components identified. It is shown how the predictive power of the different components depends on sampling frequencies.
volatility realized variance intraday seasonality volatility prediction high-frequency data tick data fractional integration sampling frequency