Improving the value at risk forecasts: Theory and evidence from

The recent financial crisis has raised numerous questions about the accuracy of value at risk (VaR) as a tool to quantify extreme losses. In this paper we develop data driven VaR approaches that are based on the principle of optimal combination and that provide robust and precise VaR forecasts for periods when they are needed most, such as the recent financial crisis. Within a comprehensive comparative study we provide the latest piece of empirical evidence on the performance of a wide range of standard VaR approaches and highlight the overall outperformance of the newly developed methods. & 2012 Elsevier B.V. All rights reserved.