ISDA SIMM at 10: A Shared Framework for Derivatives Margin

ISDA’s Standard Initial Margin Model has become shared infrastructure for derivatives markets: it gives counterparties a common way to calculate initial margin, helping contain disputes and operational strain. A decade after global margin requirements for non-cleared derivatives took effect in September 2016, more than 400 groups of entities use the model. Estimates cited by ISDA put its reach at more than 21,000 portfolios and approximately $1 trillion in initial margin collected.

The model’s scale reflects a practical choice made before launch. As Eduardo Epperlein, Managing Director at Nomura International, and Tomo Kodama, Managing Director at Bank of America Merrill Lynch, discussed on Episode 60 of ISDA’s The Swap podcast, firms could use a standard margin grid or build proprietary models. The grid did not adequately account for offsetting portfolio risks; individual models, meanwhile, could produce materially different calls for the same counterparty relationship.

That split risked making funding less predictable and disputes harder to resolve. Kodama said a market built around competing methods would be “more expensive, more uncertain, less transparent, and more operationally difficult.” Smaller firms could have found it especially hard to reproduce calculations from larger counterparties. Industry participants began raising these concerns with regulators in 2013 and later asked ISDA to coordinate a common approach.

Common calculations, shared oversight

The design brief was demanding: a model needed to reflect portfolio risk, be transparent and reproducible, and work for firms of different sizes. The framework adapted and calibrated sensitivity-based methods used in regulatory approaches to measure counterparty exposure. Its adoption also depended on the plumbing around the calculations: legal documentation, data formats, reconciliation processes and operational standards. ISDA formed committees covering those areas alongside methodology.

“Without that level of coordination and without that level of transparency, there was no way the margin rules would have been implemented without some level of market disruption,” ISDA Chief Executive Scott O’Malia said. His point gets at the model’s broader function: standardization is useful only when counterparties can exchange, check and act on the results.

SIMM’s governance is designed to keep those results under review. Epperlein described weekly industry meetings on performance and potential changes, supported by data contributed by participating firms. The group developed a shared monitoring framework and a commonly understood backtesting process, rather than relying only on separate assessments of individual firms’ models. Quarterly backtesting reports are shared with regulators to assess whether the model continues to meet its intended risk coverage.

No single mathematical model can capture every possible portfolio without becoming impractically complex. SIMM therefore includes a process for identifying portfolios that need extra margin. In the latest quarterly monitoring exercise Kodama cited, 13 of more than 11,000 portfolios examined needed remediation, through portfolio adjustments or additional collateral, to meet the 99% risk coverage standard. The figure points to a targeted backstop within a much larger system; it doesn’t mean every portfolio carries the same risk or requires the same treatment.

ISDA has made 17 changes to SIMM over its first decade. O’Malia credited that ongoing maintenance with keeping the framework relevant as markets changed. The model’s reach has also supported an ecosystem of nearly 70 vendors offering related services. Standardized risk data and exchange formats have helped these firms build tools for margin reconciliation, portfolio optimization and collateral management. Kodama described SIMM as an industry utility, rather than a system built to advantage one institution.

Margin through volatile periods

One central design aim is to limit procyclicality: a sharp jump in margin during market stress can increase firms’ immediate liquidity needs just as funding is harder to manage. Epperlein said SIMM’s calibration uses at least four years of historical data, with a stressed period making up 25% of the sample. That structure is intended to keep short bursts of volatility from driving abrupt margin increases.

Discussing the March 2020 dash for cash and the UK gilt crisis in 2022, Epperlein said SIMM performed well and required relatively limited adjustments. He contrasted its behavior with central counterparty margin models, which he said can be more procyclical. “SIM is designed to be less pro-cyclical, a bit conservative, but that’s its strength,” he said. Kodama underlined the funding trade-off: low margin in calm conditions offers little comfort if requirements surge during stress.

The update schedule has changed too. ISDA moved from annual to semiannual calibration in 2025, creating more frequent opportunities to reflect market conditions while retaining a regular timetable. Epperlein said recalibration became routine and no longer needs separate pre-approval when the underlying methodology stays the same. For firms managing collateral and liquidity, a known schedule can make model changes easier to plan around.

Where the framework may expand

Kodama identified additional jurisdictions, particularly China, as possible areas for broader use, alongside new currencies, yield curves and financial products that may need development. Tokenization and cryptocurrencies could also warrant closer attention as their use grows. He cautioned that additions need evidence: unnecessary changes risk creating inconsistencies in a framework whose value depends on common methods.

Epperlein called SIMM a “living model” that must respond as risk changes, while retaining clear governance and defined update procedures. Kodama recalled the September 2016 launch as an “emotional milestone” and described the work as a constructive contribution to the industry. The next test is whether the framework can accommodate new exposures without weakening the shared standards that have made it usable across firms.