Forecasting stock volatility using pseudo-out-of-sample information

This paper proposes a novel forecast combination method and investigates the impact of technical indicators on volatility prediction in the Chinese stock market. Firstly, our analysis reveals that technical indicators based on good and bad volatility have a stronger explanatory power on stock volatility, exhibiting an evident asymmetric effect. In addition, we introduce two new categories of technical indicators based on price skewness risk and kurtosis risk, which exhibit more robust and significant predictive power on volatility than existing indicators. Finally, we propose a new forecast combination method that employs adjusted out-of-sample R2 as a weight, which outperforms a series of existing forecasting models. These findings are robustly confirmed in multiple robust tests, demonstrating the efficacy of our proposed approach.