Updated 1 November 2026
Why SIMM versions matter
The ISDA Standard Initial Margin Model (SIMM™) is the industry's common method for calculating regulatory initial margin on non-cleared derivatives. Both counterparties run the same model on the same sensitivities and should arrive at close numbers. That only works if both sides run the same version, with the same parameters, on the same date.
ISDA recalibrates SIMM regularly so that risk weights and correlations reflect recent market data, including a period of stress. Each release changes margin amounts — sometimes materially for specific portfolios — and each one needs to be implemented, tested and rolled out in step with your counterparties and custodians. For firms that entered the uncleared margin rules in the later phases, this release cycle is now a permanent operational task.
How versions and recalibrations are organised
SIMM has a methodology version (for example v2.5, v2.6, v2.7, v2.8) and, more recently, interim recalibrations identified with a year-month suffix (for example v2.7+2412). v2.8+2512 is the current Everix default. ISDA published v2.9+2606 on 30 October 2026, effective 5 December 2026 (first exchange 7 December); Everix delivery is early access. Calibration has been semiannual since 2025. A methodology version can change the structure of the model — buckets, risk-factor definitions, treatment of particular products — while a recalibration typically updates parameters within the same structure.
- Methodology changes: new or merged buckets, changes to which risk factors exist, changes to how vega or curvature are computed, new product treatments.
- Parameter recalibrations: risk weights, intra- and inter-bucket correlations, concentration thresholds, vega risk weights and scaling factors, historical volatility ratios.
- Documentation releases: the methodology document, the parameter tables, and the unit-test packs that ISDA makes available to licensees.
The practical consequence: every release needs at least a parameter update and a regression run; a methodology release also needs code changes, and possibly CRIF mapping changes upstream.
SIMM building blocks, briefly
A quick recap helps to see where each release lands. SIMM is a sensitivity-based model. Its inputs are delta and vega sensitivities, plus base-correlation sensitivities where relevant, usually exchanged in the Common Risk Interchange Format (CRIF). Curvature margin is derived under the SIMM methodology. CRIF tags sensitivities with risk type, qualifier, bucket and label.
- Six risk classes: interest rate, credit qualifying, credit non-qualifying, equity, commodity and FX.
- Four product classes: RatesFX, Credit, Equity and Commodity. Margin is computed per product class and the product-class amounts are added, without diversification between them.
- Within a product class, each risk class produces delta, vega and curvature margin (plus base-correlation margin for credit qualifying), and the risk classes are aggregated with prescribed correlations.
- Within a risk class, weighted sensitivities are aggregated within buckets and then across buckets, with concentration factors scaling up large exposures.
- Add-ons and multipliers can apply on top, for example for products or risks not captured by sensitivities.
Every one of these layers has parameters, and a release can touch any of them. That is why a recalibration that looks small in the release notes can still move margin noticeably on a specific, concentrated or well-hedged portfolio.