Basically VaR backtesting is comparing the 1 day VaR (not the 20 day VaR) to what actually happened on the portfolio from one day to next.
The VaR Backtesting compares:
Back-testing can be done on either a Dirty or Clean Basis.
Dirty back-testing is comparing the 1 day VaR to the movement in the price of the portfolio from one-day to next. It does not consider or take account of the fact that actually the composition of the portfolio may have changed through new purchases and sales.
Clean back-testing considers yesterday’s portfolio and the re-values yesterday’s portfolio based on today’s prices. In otherwords, it strips out the impact of new purchases and sales.
From a Funds-Axis system perspective, please note that for back-testing purposes:
The “dirty” 1 day return can be calculated by the system and used for the “Portfolio 1 day return” where the client captures the daily shareclasses prices in the system. Note, the system needs also to be set-up to identify which shareclass should be used when considering price movements from one day to the next.
the clean 1 day return can be calculated by the system and used for the Portfolio 1 day return. Note, this requires daily prices to be available to the system,
An overshoot is if the portfolio actually lost more than the 1 day VaR predicted
The system automatically records an “Overshoot” where the 1 day return is less than the VaR.
For UCITS, the number of overshoots needs to be reported to regulators where it is greater than 4 overshoots in a 250 day history.
Within the reporting, you can see the history of back-testing, how many overshoot there have been together with any overrides and explanations.
Not least because of the requirement to report to Regulators, we provide clients with the ability to override an overshoot. They can also provide explanations both for overshoots and overrides.
There is audit trail and maker-checker in respect of this.