Contributions to short-term financial risk management

This thesis presents instruments and methodologies for fi nancial risk management applications: A method of estimating instantaneous volatility from transaction data is developed. It explicitly accounts for microstructure noise. Furthermore, an econometric method is introduced which copes easily with short-term patterns in time series such as the intraday volatility patterns. Regarding extreme events, important aspects of Lévy processes are discussed. A univariate approximation of Student Lévy processes is developed. In the context of multivariate Lévy processes, a modified, unbiased simulation algorithm is presented. The concept of jump tail dependence is discussed, which is a property of the Lévy copula. Especially on the short-term horizon, it is of special relevance for optimal asset allocation. Asymptotical results are derived, which allow for the estimation of jump tail dependence.