The estimation and management of risks is an important and complex task that faces market regulators and financial institutions. It has become apparent that more accurate and reliable quantitative measures of risk are needed to avert, or at least minimize, the undesirable effects on a given portfolio of large fluctuations in the conditions of the market. To accomplish this task, a series of computational tools has been designed, implemented, and incorporated into MatRisk, an integrated environment for risk assessment developed in MatLab. Besides standard measures, such as Value at Risk (VaR), the application MatRisk allows the calculation of other more sophisticated risk measures. These novel risk measures (Shortfall, MaxVaR, conditional VaR) have been introduced by a number of authors to address the inability of VaR to characterize the structure of risk properly.Amongst the extensions of the classical VaR methodology incorporated into MatRisk is the possibility of calculating percentiles for non-normal distributions (e.g., hyperbolic distributions, mixture of Gaussians, and the like), which may provide a more accurate model of the actual behavior of the portfolio returns. The application also allows the calculation of risk measures based on the distribution of extreme events, such as MaxVaR and Expected Shortfall. Finally, risk measures derived from estimates of the conditional probability distribution of returns can be obtained. To produce these conditional risk estimates, MatRisk includes extensions to carry out time analysis in terms of autoregressive models, such as ARCH, GARCH and MixGARCH (probabilistic mixtures of GARCH models).
