How does ISDA SIMM calculate initial margin for non-cleared derivatives, and what are the key risk buckets?

ISDA SIMM (Standard Initial Margin Model) is the industry-standard method for calculating initial margin on bilateral (non-cleared) OTC derivatives. It uses a sensitivity-based approach where margin depends on how the portfolio's value changes with respect to key risk factors.

Why SIMM Exists

Post-crisis regulations (BCBS-IOSCO Uncleared Margin Rules) require bilateral counterparties to exchange initial margin. Rather than each bank using its own proprietary model (creating disputes), ISDA developed SIMM as a standardized, transparent methodology that both sides compute identically.

The Calculation Framework

SIMM proceeds in four steps:

1. Compute sensitivities: For each trade, calculate delta, vega, and curvature sensitivities to risk factors in each risk class.

1. Weight sensitivities: Multiply each sensitivity by a prescribed risk weight (calibrated to a 99% VaR over 10 days).

1. Aggregate within buckets: Within each risk class, sensitivities are grouped into buckets (e.g., currency buckets for interest rate risk). Correlations within and across buckets determine how risks offset.

1. Aggregate across risk classes: The six risk class margins are combined using a formula that recognizes limited diversification benefit.

SIMM Risk Classes

Example: Birchwood Partners IM Calculation

Birchwood Partners has three non-cleared trades with Maplewood Bank:

• A 10-year USD interest rate swap: IR delta sensitivity = $4.2M per 1bp
• A EUR/USD FX forward: FX delta sensitivity = EUR 12M
• Equity options on a European index: equity vega sensitivity = $850K per 1vol

SIMM weights each sensitivity by its prescribed risk weight, aggregates within each risk class using the correlation matrix, then combines across risk classes:

> Total SIMM = sqrt(IM_IR^2 + IM_FX^2 + IM_Equity^2 + 2psiIM_IR*IM_FX + ...)

Where psi represents the cross-risk-class correlation (typically low, around 0.15-0.27).

Suppose after weighting and intra-class aggregation: IM_IR = $8.1M, IM_FX = $3.4M, IM_Equity = $2.8M. With cross-class correlations of 0.20:

Total SIMM ≈ sqrt(8.1^2 + 3.4^2 + 2.8^2 + 20.208.13.4 + 20.208.12.8 + 20.203.4*2.8)

= sqrt(65.61 + 11.56 + 7.84 + 11.02 + 9.07 + 3.81)

= sqrt(108.91) ≈ $10.4M

FRM exam tip: Know the six risk classes and that SIMM is sensitivity-based (not scenario-based). Understand that cross-risk-class diversification is limited by design. Questions may ask why SIMM produces different margin than a bank's internal model — the answer is usually that SIMM uses prescribed risk weights and correlations rather than estimated ones.

Join our FRM Part II community for more margin and collateral discussions.