Get the latest updates and news from FINCAD. Subscribe and never miss a post! 


Gearing Up for SIMM
By Tony Webb, PhD | September 30, 2015

For derivatives firms, keeping pace in a marketplace characterized by ever-stricter compliance oversight is no easy feat. The current reality is that most institutions’ risk approach is not comprehensive enough to handle the demands of today’s changing regulatory environment. As such, the majority of these firms will need to adjust and modernize their risk practices in order to successfully cope.

One major regulation that is on the horizon for derivatives dealers is the requirement to post margin on non-centrally cleared derivatives. This requirement will be phased in, starting with variation margin, and later initial margin. The latest schedule from BCBS/IOSCO (March 2015) shows the phase-in period for posting initial margin on non-centrally cleared trades starting September 2016 for entities with the largest notional amounts of derivatives, with subsequent annual deadlines through to 2020 for entities with progressively smaller amounts.

The Standard Initial Margin Model (SIMM) is a risk-based approach to calculating initial margin, being proposed by the International Swaps and Derivatives Association (ISDA), based on an earlier “Sensitivity Based Approach” from BCBS. The SIMM will be transparent and standardized, so that there is common ground to resolve disputes around margin calculations. The SIMM uses sensitivities (i.e., generalized Greeks) as inputs, requiring the first order sensitivity with respect to every risk factor, including delta and vega for each asset class. As outlined by ISDA in their December 2013 paper, Standard Initial Margin Model for Non-Cleared Derivatives, “it is imperative that a SIMM model approximates the response to shocks with a fast calculation if we are to avoid derivative price-making coming to a standstill”, and “Using a portfolio’s Greeks, instead of full re-evaluation, application of the scenario shocks to a derivatives portfolio now becomes a simple matter of multiplication and addition, and can thus be done very fast and is easily checked for errors.”

There is an incongruity, however, in that most derivatives dealers still utilize the traditional method for calculating the necessary Greeks and other sensitivities, known as “bumping.” Unfortunately, this approach is prohibitively slow; “as slow as a turtle” my colleague Russell Goyder often analogizes. When using “bumping”, the entire portfolio must still be re-valued for every risk factor to which sensitivity is required. As I see it, the need to perform full re-valuations (“bumping”) across vast and, many times, diverse books of products, is going to place major processing strain on firms’ pricing and risk models.

It may go without saying that bumping is not the wave of the future, nor is it an efficient way of meeting the ISDA mandate. There is a better way, called Algorithmic Differentiation, or AD. Firms using AD typically experience an acceleration in risk run-time of 100-1000 times that of bumping. Where typical risk runs are often an overnight activity and limited to looking at a subset of total exposure, AD enables you to view the entire risk landscape on a pre-trade basis. Instead of spending a night peering into your risk and hoping you looked in the right places, utilizing AD enables you to uncover each and every exposure – at a fraction of the cost of bumping.

But benefits of this method are not limited to risk run-time acceleration alone. While bumping is an approximation, AD gives you exact, analytic Greeks and other sensitivities. Furthermore, as an estimative technique, bumping leaves you with a lot of opened questions. Often you’re faced with deciding how big a bump should be, whether a basis point is sufficient and whether it should be relative or absolute. With AD, this uncertainty is eliminated, as you are able to determine exact, analytic sensitivity. Furthermore, because there is no need to go back and fine-tune calculations using AD, quants are freed up to focus on more strategic work. 

FINCAD has been a pioneer in the area of AD. We’ve incorporated our own patented implementation of the methodology, known as Universal Algorithmic Differentiation™ (UAD) into F3 Platform, which was launched in 2010. UAD guarantees fast analytic exposure calculations for all valuations, from vanilla to exotic, single-trade to portfolio, under all models and valuation methodologies. Its generic architecture makes this possible, which replaces traditional model or trade-specific approaches. According to Amrish Ganatra, Founding Partner of the investment technology consulting firm, Cerebra, testing showed that UAD's speed was 2,000 times faster than bumping, while "the results were practically identical."

For more information on how UAD can help you meet compliance requirements like SIMM, while attaining competitive advantage, download our ebook: Improve Trading Performance with Algorithmic Differentiation.

About the author
Tony Webb PhD
Tony Webb, PhD
Quantitative Advisor | FINCAD

Tony Webb is an experienced manager and quantitative analyst at FINCAD having held various roles, including Director of Analytics, VP R&D, Product Manager, and manager of technical Pre-Sales Analysts in NYC. He holds an MA in Mathematics from Cambridge University, a PhD in computational fluid dynamics from UBC, and an MBA with a specialization in finance from UBC. He is currently acting as a Quantitative Advisor within the Client Services department.