Recently, the great fluctuation of the crude oil market has caused adverse effects on the economic environment, drawing more attention to accurate forecasting from researchers, investors, and policymakers. The traditional risk measurement methods mainly focus on the 1-day horizon Value at Risk (VaR), which is not enough to warn investors. And the Basel Accord III laid new emphasis on Expected Shortfall (ES) in 2019, making new requirements for risk management. In order to improve the accuracy in low-frequency risk forecast and take ES into consideration, we adopt a new method based on the Mixed Data Sampling (MIDAS) framework to jointly forecast the VaR and ES in the crude oil market through the Asymmetric Laplace density, named AL-MIDAS. It makes full use of the information contained in daily data to make direct low-frequency risk predictions, thus improving the accuracy of the predictions. And we take multi-day VaR and ES of WTI as the target and take WTI and USD index return as the independent variables separately to study the different impact of own historical data and influencing factors’ data on the forecast. The results show that, based on the AL-MIDAS, it is a good performance with WTI own historical return; while the forecasting performance with USD index daily return is unstable. What's more, compared with the historical simulation method, the performance of the new method is better. Therefore, MIDAS should be used in risk management to improve management ability.

MIxed Data Sampling,Value at Riak,expected shortfall,joint elicitable,crude oil market