Forecasting Realized Volatility of the Oil Future Prices Via Machine Learning

This paper explores the possibility of the potential usage of machine learning models in the field of realized volatility forecasting of crude oil with a vast variety of empirical analyses and robustness checks. Although the conventional heterogeneous autoregressive (HAR) model is widely accepted, it failed in out-of-sample recently. However, machine learning models with the HAR factors can improve forecasting performance significantly. The model confidence set test, Diebold-Mariano test, and Pesaran-Timmermann test support the empirical results statistically