Sample Frequency Robustness and Accuracy in Forecasting Value at Risk for Brent Crude Oil Futures

In this paper we propose a simple one-factor quantile regression model based on realized volatility to forecast Value-at-Risk (VaR). The model only uses daily realized volatility as input and thus simplifies estimation substantially compared with most other methodologies currently used to forecast VaR. We evaluate the performance of the proposed one-factor model across various sampling frequencies used to calculate realized volatility using formal coverage tests for out-of-sample forecasts and applying it to ICE crude oil futures. The results show that our one factor model performs exceptionally well for most sampling frequencies used to calculate realized volatility. In comparison with the well-known Heterogenous Auto-regressive Model of Realized Volatility (HAR-RV) and a quantile regression version of the HAR model (HAR-QREG), we find that our proposed one-factor model is much less sensitive to the sampling frequency used tocalculate realized volatility