Capturing fat tails with full re-pricing
While all assets are likely to exhibit “fat tails” in their return distributions, where extreme observations occur more frequently than a normal distribution would suggest was likely, this can be particularly acute with many derivatives, especially where they have a degree of optionality.
To properly capture this non-normal behaviour when estimating the Value at Risk of a portfolio, it is necessary for VaR to be estimated by simulation rather than parametrically and, in each simulation, the derivative should be re-priced to take account of, in particular, the change in the delta of the derivative to its underlying as the underlying asset is assumed to have changed in price.
In order to achieve this, when calculating the VaR of a portfolio that contains derivatives the EMA system uses pricing models to estimate the return of a derivative based on the estimated price of the underlying asset within each simulation. As the pricing of some derivatives is a non-linear function of the prices of the underlying assets this approach ensures that the distribution is calculated correctly. Where a portfolio has material exposure to derivatives of this type the overall distribution will be especially non-normal with very fat or thin tails as appropriate.
Not only does the EMA system estimate a realistic VaR it also calculates all the Greeks and reports on them at asset and at category level.
Combined with EMA’s FASTVaR forward looking volatility forecasts this approach enables the manager to both comply with the EU’s UCITS legislation and respond quickly to changes in portfolio risk profile.