Measuring and managing market, credit and Op risk [26]

Go to: Summary | Previous | Next   
Bullet points include: Estimate appropriate parametric loss distribution and set VaR equal to (1-alpha)-quantile E.g. multi-variate Normal (i.e. Gaussian) benchmark case Time series of, say, daily returns (/losses) on the portfolio, might assume i.i.d. Normal VaR then estimated from the relevant (1-alpha)- quantile, i.e. (where N-1(x) is inverse Normal distribution):

Contents | Prev | Next | ERM Lecture Series

Desktop view | Switch to Mobile