Topic: Contributions on quantile regression in finance and economics Title: Improving accuracy of value at risk estimation using intra-day data Authors:  Xiaochun Meng - University of Sussex (United Kingdom) [presenting]
James Taylor - University of Oxford (United Kingdom)
Abstract: Some novel VaR approaches are proposed that utilize intra-day data and overnight night return. These approaches incorporate lagged intra-day range, daily high, daily low and overnight return. A new parameter estimation procedure based on quantile regression is proposed and used. The proposed approaches are very flexible and easy to implement. An empirical analysis is conducted on five market indices. The performances of the proposed models, including a comparison with the established benchmark models, are examined using three standard VaR back-testing methods. The post-sample results show that the proposed models successfully capture the main characteristics of the financial returns and perform very competitively. One of the proposed models utilizing both the intra-day data and the overnight return is shown to forecast VaR more accurately than other models.