With the revision of the Basel II market risk framework and implementation of Fundamental Review of the Trading Book (FRTB) regulation (coming into effect on 1 January 2019) as part of Basel III, use of Adjoint Algorithmic Differentiation (AAD) for speeding up regulatory capital calculations is gaining global support. It is the new normal: massive calculations are expected to become routine under the new FRTB rules. In addition, banks are obliged to overhaul both their market risk infrastructure and their technology capabilities.
AAD-based methodologies offer improved computational capacity, making it easier to comply with the new requirements.
The best available technique
AAD, fast becoming the centerpiece of computational finance, seems to be the best available technique: almost no manual effort is required to calculate sensitivities in FRTB and FRTB-CVA.
FRTB and the Credit Valuation Adjustment (CVA) requirements—the FRTB-CVA framework introduced with the Basel Committee on Banking Supervision Document 352 (BCBS 352)—will give banks an opportunity to opt for a sensitivities-based standardized approach (SA) to calculating their regulatory capital.
In its explanatory note, the Basel Committee on Banking Supervision outlines elements of the revised market risk framework including the revised SA. The latter provides a “sensitivities-based method for capturing three risk sensitivities, namely, delta, vega, and curvature risks”. Methods to define the sensitivities can be found in the FRTB documents. The FRTB-CVA documents offer similar definitions.
According to Alexander Sokol, CompatibL’s CEO, AAD is particularly advantageous for calculations in FRTB-CVA, where the CVA figure depends on hundreds or even thousands of curves for netting sets. Such dependency complicates the process even with the reduced number of buckets per curve: in FRTB-CVA, unlike FRTB, the total number of sensitivities exceeds 1,000, on average. With AAD, the calculation can be performed 100 times faster.
ISDA SIMM methodology promotes AAD
In September 2016, the International Swaps and Derivatives Association (ISDA) announced the launch of the ISDA Standard Initial Margin Model (SIMMᵀᴹ), created in response to the new initial margin calculation requirements.
Many definitions for sensitivities offered by the ISDA SIMMᵀᴹ are similar to those from the FRTB banking regulation. However, the model adds a few alternative definitions: it allows the use of the central or backward difference methods, or of a smaller shock size and scale-up. This alternative is fully compatible with AAD.
Once finalized, BCBS 352 may not be amended. However, with the rapidly growing support for AAD affording immense potential for calculations under FRTB final rules, the restriction may be further addressed in subsequent frequently asked questions and/or technical guidance issued by country regulators.
The FRTB-CVA is still a draft document. The Basel Committee still has a chance to clarify its definitions and adapt its list to include the SIMM definitions.
Since 1985, the International Swaps and Derivatives Association has worked to make the global derivatives markets safer and more efficient.
ISDA’s work in three key areas—reducing counterparty credit risk, increasing transparency, and improving the industry’s operational infrastructure—shows the strong commitment of the Association toward its primary goals: to build robust, stable financial markets and a strong financial regulatory framework.