What Is ISDA SIMM and How Is Initial Margin Calculated?

The ISDA Standard Initial Margin Model (SIMM) is a sensitivity-based methodology that two parties to an uncleared derivatives trade use to calculate how much initial margin each must post to the other. Developed by the International Swaps and Derivatives Association after the 2008 financial crisis, it replaced a patchwork of private margin arrangements with a single transparent formula. The current production version is SIMM v2.8+2506, effective since December 2025, with parameter updates now occurring twice a year.1International Swaps and Derivatives Association. ISDA Publishes ISDA SIMM Methodology, Version 2.8+2506 The calculation converts each position’s risk sensitivities into a margin figure calibrated to cover a one-in-a-hundred loss over a ten-business-day close-out window.

The Regulatory Standard the Number Must Meet

Everything in the model exists to satisfy one requirement. Initial margin must be sufficient to cover losses at a one-tailed 99 percent confidence interval over a ten-business-day holding period.2eCFR. 17 CFR 23.154 – Calculation of Initial Margin Put plainly: the margin held should absorb losses in all but the most extreme one percent of ten-day market moves, on the assumption that a defaulted counterparty’s positions take ten days to close out and replace. The calibration data used to build the model must include a period of significant financial stress.3Bank for International Settlements. Margin Requirements for Non-Centrally Cleared Derivatives

This is where SIMM differs from a full portfolio simulation. Rather than running thousands of scenarios on every trade, the model takes standardized risk sensitivities as inputs and applies pre-calibrated risk weights and correlations to produce a margin number that meets the same statistical target.

The Three Risk Sensitivities

For every position in scope, a firm computes three sensitivities:

  • Delta: how much the position’s value changes when the underlying price moves by a small amount.
  • Vega: how much the position’s value changes when implied volatility shifts.
  • Curvature: the non-linear price behavior that delta alone doesn’t capture, particularly important for options.

These sensitivities come from the firm’s own pricing engines. Their accuracy sets a ceiling on the accuracy of the margin figure. Poorly calibrated pricing models either over-collateralize, tying up capital, or under-collateralize, creating regulatory exposure.

The Six Risk Classes

Every derivative is assigned to one of six risk classes:4International Swaps and Derivatives Association. ISDA SIMM Methodology

  • Interest Rate: swaps, swaptions, and other rate-linked products.
  • Qualifying Credit: higher-liquidity debt instruments such as investment-grade bonds and related derivatives.
  • Non-Qualifying Credit: more complex or lower-liquidity credit obligations.
  • Equity: derivatives on individual stocks, indices, or equity volatility.
  • Commodity: derivatives on physical goods like oil, metals, and agricultural products.
  • Foreign Exchange: currency forwards, options, and cross-currency swaps.

The class assignment matters because netting is allowed inside a risk class but never across classes. Offsetting commodity positions can reduce the margin figure for the commodity class, but gains in commodities cannot offset losses in equities. The final portfolio margin is the sum of the six class-level results. This deliberate non-diversification is what keeps the total conservative.

How the Model Aggregates Sensitivities

Within each risk class, sensitivities are grouped into risk buckets. Interest rate sensitivities, for example, are bucketed by currency and by tenor. The model then applies standardized risk weights to each bucketed sensitivity and standardized correlation parameters to combine buckets into a single margin amount for that class. The correlations determine how much diversification credit a firm receives for holding positions that tend to move in opposite directions inside the class.

The four ingredients that turn raw sensitivities into a margin number are therefore: the bucket assignment, the risk weight applied to each bucketed sensitivity, the correlation parameters used to combine them, and any concentration adjustment. ISDA supplies all four; the firm supplies the sensitivities.

Concentration Risk

When a firm holds an unusually large position relative to the liquidity of a particular market, the model applies a concentration factor that scales the margin up. If the sum of a firm’s net sensitivities in a given bucket exceeds the concentration threshold defined for that bucket, the weighted sensitivity is multiplied by a factor greater than one. The larger the position relative to the threshold, the higher the multiplier. Outsized exposures in illiquid corners of the market therefore attract proportionally higher margin, not just linearly higher margin.

The Inputs the Calculation Requires

All sensitivity data must be formatted in the Common Risk Interchange Format (CRIF), a standardized column-based file layout that ISDA developed so any two counterparties can exchange risk data in a consistent way.5International Swaps and Derivatives Association. ISDA Common Risk Interchange Format Every trade must be mapped to the correct SIMM risk bucket, and the file must be generated frequently enough to support daily margin calls.

Standardization here serves a specific purpose. When both sides describe their portfolios using the same template, calculation disputes become easier to isolate. Firms rarely share the full underlying sensitivity data because it is proprietary, but comparing SIMM results at the bucket level usually reveals where two calculations diverge.

Any firm running the model — including vendors offering SIMM-based services and end users calculating on their own trades — must hold a license from ISDA.6International Swaps and Derivatives Association. ISDA SIMM Licensing FAQ

Semiannual Recalibration

Starting in 2025, ISDA moved from annual to semiannual recalibration of SIMM parameters. The primary calibration occurs in the first half of each year, with updated parameters taking effect in August, and reviews all SIMM parameters. A secondary calibration runs in the second half, with changes effective in February, focused on the main delta risk weights.7International Swaps and Derivatives Association. ISDA SIMM to Move to Semiannual Calibration Each cycle runs about seven and a half months including the regulatory notification period. The old process combined an annual calibration with quarterly checks that could trigger ad hoc updates after a new stress event.

Who Has to Use the Model

Falling within scope depends on a single metric: the Average Aggregate Notional Amount (AANA) of uncleared derivatives held by the firm and its corporate group. Under CFTC rules, “material swaps exposure” exists when a firm’s average month-end aggregate notional amount of uncleared swaps, uncleared security-based swaps, foreign exchange forwards, and foreign exchange swaps exceeds $8 billion.8GovInfo. 17 CFR 23.151 – Definitions The equivalent threshold in the EU and many other jurisdictions is €8 billion.9International Swaps and Derivatives Association. Countdown to Phase 6 Initial Margin

The measurement window is three specific months. For the 2026 compliance year, firms use the month-end aggregate notional amounts from March, April, and May 2026.10International Swaps and Derivatives Association. OTC Derivatives Compliance Calendar Only the last business day of each month counts, and a swap between a firm and a margin affiliate is counted once. Regulators specifically prohibit structuring activity to duck below the threshold at month-end.8GovInfo. 17 CFR 23.151 – Definitions Firms must recalculate AANA annually, since crossing the line in either direction changes their obligations.

The $50 Million Relationship Threshold

Exceeding the AANA threshold does not mean a firm must post collateral the next morning. There is a second, relationship-level threshold. The initial margin threshold amount under CFTC rules is $50 million in aggregate credit exposure across all uncleared swaps between two corporate groups.8GovInfo. 17 CFR 23.151 – Definitions If the calculated bilateral initial margin between two groups stays below $50 million, no actual collateral moves for that relationship. The documentation still needs to be in place so margin can flow immediately once exposure crosses the line.

One boundary worth flagging: the model applies to initial margin, which covers potential future exposure and moves both ways on a gross basis. Variation margin, which covers current mark-to-market changes and flows one way, is a separate obligation that has been in place for covered entities since March 2017 and is not calculated using SIMM.

Regulatory Approval and Ongoing Validation

Using SIMM requires written approval from the CFTC or the relevant prudential regulator, and keeping that approval is an ongoing obligation. A firm must notify the Commission in writing at least 60 days before extending the model to a new product type, making a material change to the model, or changing material modeling assumptions.2eCFR. 17 CFR 23.154 – Calculation of Initial Margin Approval can be rescinded if the model no longer meets the statutory standard.

Firms must backtest the model, comparing predicted margin against actual profit and loss outcomes. If it consistently underestimates risk, the firm may need to hold additional capital or revise its internal valuation methods. Benchmarking against alternative risk measurement approaches is also expected. Both activities should be run by personnel independent of the teams that built or operate the model, and firms should keep detailed records of every validation, backtest, and model change. Losing the ability to use SIMM forces a reversion to the standardized margin schedule, which is far more punitive for most portfolios.