The impact of trading volume, number of trades and overnight returns on forecasting the daily realized range

Intraday data of 26 German stocks are used to investigate whether the information contained in trading volume and number of trades as well as in various specifications of overnight returns can improve one-step-ahead volatility forecasts. For this purpose, a HAR model of the realized range adjusted for discrete trading is augmented by each of these variables and compared with the model's default form. The results show that the considered liquidity measures lead to very modest improvements in forecasting performance. The overnight returns exhibit some in-sample forecasting power. However, the accuracy improvement of out-of-sample forecasts is unequivocally non-significant.