Quantitative standardsQuantitative standards

Quantitative standards

An ADI with approval to use an internal model must comply with the quantitative criteria outlined in paragraphs 29 to 34 of this Attachment for the purpose of calculating its capital charge. This does not preclude an ADI from imposing more stringent criteria if it wishes to do so.
VaR must be calculated on a daily basis, using a 99 per cent, one-tailed confidence interval and a 10-day holding period. An ADI which uses VaR numbers calculated according to a shorter holding period scaled up to ten days must justify the reasonableness of its approach to APRA’s satisfaction
The historical observation period (sample period) chosen for calculating VaR must have a minimum length of one year. An ADI using a weighting scheme or other method for the historical observation period cannot have a weighted-average time lag of the individual observations of less than six months[\[43\]](#_ftn43). APRA may require an ADI to calculate its VaR using a shorter observation period if APRA’s considers this is justified by a significant upsurge in price volatility.
An ADI must update its data sets at least monthly and must reassess them whenever market prices are subject to material changes. An ADI must have processes in place to update their data sets more frequently.
An ADI may only recognise empirical correlations within and across broad risk categories if approved in writing by APRA
An ADI’s model must accurately capture the unique risks associated with options within each of the broad risk categories, in particular the non-linear price characteristics of option positions.
The stressed VaR measure must be calculated using a 10-day, 99th percentile, one-tailed confidence interval value-at-risk measure of the current portfolio, with VaR model inputs calibrated to historical data from a continuous 12-month period of significant financial stress relevant to the ADI’s portfolio. The choice of historical period is subject to APRA approval, and the ADI must review the appropriateness of the choice as part of its regular model validation process. This stressed VaR must be calculated at least weekly.