(C)VaR estimates for different distribution classes

  • Andreev, Andriy (Projektledare, akademisk)

Projekt: Externt finansierat projekt



This study suggests algorithmic procedures for calculation of value-at-risk and conditional-value-at-risk using Pearson's family of distributions. This family contains most of classical distributions, including heavy tailed ones, like t-distribution, etc. It has been noticed that the best fitting t-distributions for financial data often assume non-integer degrees of freedom. Calculation of risk measures in this case can be simplified by incorporating a simple analytical relation between degrees of freedom and kurtosis, suggesting a correction term in calculation of VaR. We extend these findings on Pearson class of distributions. Suggested procedures are streamlined for risk management applications, and, using Mathematica package,- presented in analytical form, wherever possible. Numerical work has been carried out by S-Plus and Mathematica computing packages. Examples of calculations for several financial time series are also included.
Gällande start-/slutdatum09.01.200301.01.2008


Utforska forskningsämnen som berörs av detta projekt. Dessa etiketter genereras baserat på underliggande ansökningar/anslag. Tillsammans bildar de ett unikt fingeravtryck.