Extreme Risk Forecast for Quantitative Financial Risk Management

Abstract

Value at Risk (VaR) is one of the key risk measures for quantitative financial risk management. VaR measures extreme risk, which has a small probability but a significant consequence to financial institutions. This thesis develops methods based on an extended extreme value approach to improve the forecast skill of VaR. The proposed methods improve the forecasting accuracy, robustness, efficiency and outperform the existing methods in the literature.