# ISDA SIMM Methodology: Delta, Vega & Curvature Margin Guide

> Calculate ISDA SIMM Initial Margin: six additive risk classes, 10-day 99% Delta/Vega/Curvature formulas, concentration thresholds, and annual calibration.

Canonical: https://finlexicon.org/wiki/isda-simm-v2-6-delta-vega-curvature-risk-weights-correlations
Last updated: 2026-11-14T19:37:41.205Z

**Publication:** FinLexicon Wiki
**Category:** Isda Csa Derivatives Collateral
**Published:** 2026-11-13T23:10:25.915Z
**Author:** Arthur Pendleton, CPA, JD (CPA, JD in International Banking & Basel Regulation)
**Reviewer:** Not recorded

## Answer-first summary

Calibrated to a 10-day 99% confidence interval including a 25%+ stress period, ISDA SIMM aggregates analytical Delta, Vega, and Curvature sensitivities across six asset classes without cross-asset correlation offsets.

## Key takeaways

- ISDA SIMM sums Initial Margin strictly additively across six Risk Classes (IR, Credit Qualifying, Credit Non-Qualifying, Equity, Commodity, FX) with zero cross-asset offset.
- Within each Risk Class, SIMM combines first-order Delta, implied-volatility Vega, and Taylor-scaled second-order Curvature (Gamma) charges plus Concentration Risk Factors (CR_b >= 1).
- SIMM’s "3+1" calibration weights 75% recent 3-year market volatility and 25% 2008–2009 stress volatility, preventing procyclical margin collapses during quiet regimes.
- Counterintuitively, on a 100% one-way directional swap book (NGR = 1.0), Regulatory Grid Schedule IM (1%–4% notional) can actually be *lower* than ISDA SIMM—though firms cannot cherry-pick!

## Related financial entities

- **ISDA 2002 Master Agreement & Credit Support Annex (CSA) Close-Out Netting** (ISDA-2002-CSA) — license: ISDA 2002 Master Agreement & Credit Support Annex (CSA) Close-Out Netting (ISDA-2002-CSA) refers to an institutional financial benchmark and regulatory standard monitored by International Swaps and Derivatives Association (INTL), currently calibrated at 0.0% (Netted Exposure Haircut) as of October 2026.
- **ISDA 2002 Master Agreement & Credit Support Annex (CSA) Close-Out Netting — 2026 Standard** (ISDA-2002-CSA-STD) — license: ISDA-2002-CSA 2026 Institutional Benchmark Framework is defined as the standardized quantitative and compliance methodology published by International Swaps and Derivatives Association governing ISDA 2002 Master Agreement & Credit Support Annex (CSA) Close-Out Netting across INTL capital markets (0.0% Netted Exposure Haircut, October 2026).
- **Standardized Approach for Counterparty Credit Risk (SA-CCR) Alpha Factor** (SA-CCR-EAD) — license: Standardized Approach for Counterparty Credit Risk (SA-CCR) Alpha Factor (SA-CCR-EAD) refers to an institutional financial benchmark and regulatory standard monitored by BIS BCBS CRE52 (INTL), currently calibrated at 1.4x (Supervisory Alpha Scalar) as of October 2026.

## Analysis

How do two competing derivatives dealers—each running proprietary pricing libraries—agree to the penny every morning at 08:30 on a **10-day 99% confidence Initial Margin** figure for 8,000 uncleared swaps? The answer is the **ISDA Standard Initial Margin Model (ISDA SIMM)**, structurally mirrored on Basel’s [FRTB Sensitivities-Based Method](/wiki/frtb-fundamental-review-trading-book-sensitivities-sbm-pla) and governed under the ISDA SIMM Governance Framework (`v2.6` / `v2.7` annual recalibrations) with supervisory non-objection from the [Federal Reserve](/regulators/u-s-federal-reserve-system-156), CFTC, [ECB](/regulators/european-central-bank-160)/EBA, and [UK PRA](/regulators/prudential-regulation-authority-uk-173).

## Six Siloed Risk Classes With Zero Cross-Asset Diversification

Per BCBS-IOSCO Margin Requirement Principle 3.1—as embedded in our [2002 ISDA Master Agreement & SIMM cornerstone guide](/wiki/isda-2002-master-agreement-vm-im-csa-simm-netting-guide)—regulatory Initial Margin models **may recognize hedging and diversification within a single asset class, NEVER across distinct asset classes**! Accordingly, ISDA SIMM partitions every trade’s risk factors into **six strictly additive Risk Classes** (plus regulatory notional/multiplicative add-ons where required):

`SIMM_Total = SIMM_IR + SIMM_CreditQualifying + SIMM_CreditNonQualifying + SIMM_Equity + SIMM_Commodity + SIMM_FX`

Notice how multi-asset hybrids map cleanly: a **USD/EUR Cross-Currency Basis Swap** generates `IR` sensitivities in both the USD and EUR interest-rate buckets **plus** an `FX` sensitivity on spot principal—entering *both* `SIMM_IR` and `SIMM_FX`!

![ISDA SIMM v2.6 four-step calculation pipeline from CRIF Greeks to within-bucket/cross-bucket variance-covariance and 6-class sum](/api/figure?type=flow&title=ISDA%20SIMM%20v2.6%20%2F%20v2.7%20Analytical%20Margin%20Aggregation%20Engine&subtitle=Transforming%20CRIF%20Greeks%20into%2010-day%2099%25%20non-procyclical%20Initial%20Margin&steps=1.%20CRIF%20Sensitivities%3A%3ACompute%20standardized%201bp%20IR%2FCredit%20DV01%2C%201%25%20Eq%2FFX%2FCmdty%20Delta%20%26%20Vega%20per%20trade%7C2.%20Concentration%20CR%3A%3ACompare%20bucket%20sum%20vs%20Threshold%20T_b%3B%20scale%20WS_k%20%3D%20RW_k%20*%20s_k%20*%20max%281%2C%20sqrt%28%20s%20%2FT_b%29%29%7C3.%20Bucket%20%26%20Class%20Agg%3A%3AAggregate%20intra-bucket%20K_b%20via%20rho_kl%2C%20then%20cross-bucket%20via%20gamma_bc%3B%20add%20Vega%20%2B%20Curvature%7C4.%20Additive%206-Class%20Sum%3A%3ASum%20IR%20%2B%20CreditQ%20%2B%20CreditNonQ%20%2B%20Equity%20%2B%20Commodity%20%2B%20FX%20%28zero%20cross-class%20offset!%29&metrics=Confidence%20Target%3A%3A99%25%20%2F%2010-Day%3A%3AOne-Tailed%20Close-Out%20Tail%7CCalibration%20Mix%3A%3A75%25%203Y%20%2B%2025%25%20St.%3A%3AAnti-Procyclical%20Anchor%7CIR%20Tenor%20Vertices%3A%3A12%20Tenors%3A%3A2w%2C%201m%2C%203m%20...%2020y%2C%2030y%7CCross-Class%20Corr.%3A%3A0%25%20Offset%3A%3AStrictly%20Additive%206%20Classes)

## Four Margin Components Inside Each Risk Class: Delta, Vega, Curvature, and Base Correlation

Within each Risk Class `X`, SIMM sums four standardized analytical Greek charges (`SIMM_X = DeltaMargin_X + VegaMargin_X + CurvatureMargin_X + BaseCorrMargin_CreditQ`):

1. **Delta Margin (`First-Order Linear Risk`)**: Captures parallel shifts, curve twists across **12 interest-rate tenors** (`2w, 1m, 3m, 6m, 1y, 2y, 3y, 5y, 10y, 15y, 20y, 30y`) and sub-curves (`OIS, Libor/Term-RFR, Prime, Municipal, Inflation, XCcy Basis`), equity sector buckets (`1–12`), credit rating/sector buckets, and FX spot.
2. **Vega Margin (`Implied Volatility Risk`)**: Captures `sigma * dV/dsigma` volatility-surface shifts on swaptions, caps/floors, FX options, and equity/commodity options.
3. **Curvature Margin (`Second-Order Gamma Convexity`)**: Crucial for **short-option and digital/barrier books**! SIMM re-approximates 10-day option convexity using a scaled Taylor-style Vega-to-Gamma conversion (`CVR_{ik} = Sum_j SF(t_{kj}) * sigma_{kj} * dV/dsigma_{kj}`) with an asymmetric loss scaling factor `theta` so net short gamma bears full tail-loss margin while net long gamma relief is conservatively capped!
4. **Concentration Risk Factors (`CR_b >= 1.0`)**: Whenever a portfolio’s net DV01 or Vega in a single currency, issuer, or equity ticker exceeds ISDA’s daily market-depth **Concentration Threshold (`T_b`)**, SIMM scales up that risk weight by `sqrt( |Sum WS| / T_b )` AND dampens cross-bucket correlations—mirroring longer >10-day liquidation horizons!

## Quarterly Backtesting, Annual Recalibration (v2.6 / v2.7), and 3+1 Stress Period Weighting

To prevent procyclical margin cliffs yet adapt when regime shifts occur (such as the 2022 global sovereign rate volatility surge or the September 2022 UK Gilt LDI shock), ISDA runs **quarterly industry backtesting** across dealers like [Bank of America](/institutions/bofa), [BNP Paribas](/institutions/bnp-paribas), and [Bank of Montreal](/institutions/bmo).

Every SIMM parameter table (`Risk Weights`, intra-bucket correlations `rho`, inter-bucket correlations `gamma`, and concentration thresholds) is calibrated on a **"3+1" historical sample**: **75% weight on the latest rolling 3 years** of market returns plus **25% weight on a fixed high-volatility Stress Period** (`2008–2009` crisis window), preventing SIMM risk weights from collapsing during calm bull markets!

| ISDA SIMM Risk Class | Sensitivity Definition (`s_k`) | Bucket Granularity | Illustrative Risk Weight Range | Concentration Threshold (`T_b`) Basis |
| --- | --- | --- | --- | --- |
| Interest Rate (`IR`) | PV change for `+1 bp` (`0.0001`) curve shift | Per Currency (Reg, Low-Vol JPY, High-Vol) x 12 Tenors | ~48–125 bps (Regular CCY 2y–30y) | ~$210M–$310M/bp USD/EUR vs lower EM |
| Credit Qualifying (`CreditQ`) | PV change for `+1 bp` issuer CS01 shift | 12 Rating x Sector buckets (IG 1–6, HY/NR 7–12) | ~45–480 bps across IG to HY | Higher for Sovereign/Index vs Single-Name |
| Equity (`Equity`) | PV change for `+1%` (`0.01`) spot price jump | 12 Buckets (EM vs Developed x Sector + Indexes) | ~14% (Broad Index) to ~28%–35% (EM/Small) | Market-cap & ADV tier ($M per 1%) |
| Commodity (`Commodity`) | PV change for `+1%` commodity forward shift | 17 Buckets (Coal, Crude, US Gas, EU Gas, Power, Metals, Ags) | ~16% (Gold) to ~45%–65% (EU Gas/Power!) | Physical futures open-interest liquidity |
| Foreign Exchange (`FX`) | PV change for `+1%` FX spot vs Calc CCY | High-Liquidity (USD/EUR/JPY/GBP) vs Other CCYs | ~7.2%–8.5% (G10) to ~13%–22% (EM/Volatile) | Daily BIS FX turnover tiers |

## Common Sensitivity Unit Reconciliation (CRIF File Format)

Operationally, counterparties exchange daily **Common Risk Interchange Format (CRIF)** `.tsv` files via Acadia (AcadiaSoft) IM Exposure Manager—allowing automated pinpointing of any trade-level DV01 or Vega discrepancy during [margin dispute resolution](/wiki/portfolio-reconciliation-margin-dispute-aging-emir-cftc-23-502).

![Comparison of ISDA SIMM vs Grid Schedule Initial Margin on four canonical institutional swap portfolios](/api/figure?type=bars&title=ISDA%20SIMM%20vs%20Regulatory%20Grid%20Schedule%20IM%20%28%24M%20Pre-Threshold%29&subtitle=Demonstrating%20sensitivity-based%20hedging%20recognition%20on%20%2410B%20Gross%20Notional%20Portfolios&steps=Directional%2010Y%20IRS%3A%3A%2410B%20unhedged%2010Y%20USD%20SOFR%20Swap%20%28%248.5M%2Fbp%20DV01%29%3A%20SIMM%20%3D%20%24544M%20vs%20Grid%20%284%25%29%20%3D%20%24400M!%7C2s10s%20Curve%20Steepener%3A%3A%248B%202Y%20vs%20%242B%2010Y%20DV01-neutral%20curve%20trade%3A%20SIMM%20%3D%20%24112M%20vs%20Grid%20%3D%20%24260M%20%28-57%25%20saved%29%7C95%25-Hedged%20Dealer%20Book%3A%3A%2410B%20client%2Fhedge%20IRS%20pairs%20%28NGR%20%3D%200.10%29%3A%20SIMM%20%3D%20%2438M%20%28%3C%20%2450M%20TH!%29%20vs%20Grid%20%3D%20%24176M!%7CShort-Gamma%20Swaption%3A%3ASold%20OTM%20swaptions%3A%20SIMM%20Curvature%20Margin%20adds%20%2B%2464M%20tail%20convexity%20protection&metrics=Hedged%20Book%20SIMM%3A%3A%2438.0M%3A%3ABelow%20%2450M%20Threshold%20%28%240%20Post!%29%7CHedged%20Book%20Grid%3A%3A%24176.0M%3A%3A4.6x%20Higher%20Margin%20Drag%7CUnhedged%2010Y%20IRS%3A%3AGrid%20Cheaper!%3A%3A%24400M%20Grid%20vs%20%24544M%20SIMM%7CCRIF%20Recon%20SLA%3A%3ADaily%20T%2B0%3A%3AAcadia%20Automated%20Matching)

## Benchmarking SIMM Efficiency Against Schedule Grid & Multilateral Compression

On balanced client-facing books, ISDA SIMM reduces Initial Margin by **65% to 88%** compared to [Regulatory Grid Schedule IM](/wiki/regulatory-grid-schedule-im-vs-isda-simm-ngr-netting-offset), and can be optimized further via [TriOptima/Quantile SIMM compression cycles](/wiki/multilateral-portfolio-compression-trioptima-simm-sa-ccr-relief) across the [2002 ISDA Master Agreement, CSA & SIMM topic hub](/topics/isda-csa-derivatives-collateral).

```formula
CR_b = max( 1.0 , sqrt( | Sum_k s_k | / T_b ) ) [ Concentration Risk Factor ]
WS_k = RW_k * s_k * CR_b [ Weighted Sensitivity ]
K_b = sqrt( Sum_k WS_k^2 + Sum_{k != l} phi_ij * rho_kl * WS_k * WS_l )
DeltaMargin = sqrt( Sum_b K_b^2 + Sum_{b != c} gamma_bc * g_bc * S_b * S_c ) + K_residual

Worked 2s10s USD SOFR Curve Steepener SIMM Audit (+$1.50M/bp 2Y DV01 [RW_2Y = 61 bps]; -$1.50M/bp 10Y DV01 [RW_10Y = 64 bps]; rho_{2Y,10Y} = 0.68):
 Net USD Sum |s_k| = |$1.50M - $1.50M| = $0 << $310M/bp USD T_b => CR_USD = 1.00
 WS_2Y = 61 * (+$1.50M) = +$91.50M | WS_10Y = 64 * (-$1.50M) = -$96.00M
 K_USD = sqrt( 91.50^2 + (-96.00)^2 + 2 * 0.68 * (91.50) * (-96.00) ) = sqrt( 8372.25 + 9216.00 - 11946.24 ) = sqrt( 5642.01 ) = $75.11M!
```

## Frequently asked questions

### Can a hedge fund or bank switch freely back and forth between ISDA SIMM and Regulatory Grid Schedule whenever one is lower?

Strictly prohibited ("no cherry-picking") under BCBS-IOSCO Principle 3.2, EMIR, and U.S. UMR rules! Counterparties must commit consistently to SIMM or Grid per asset class unless a fundamental change in business model occurs (subject to regulatory governance/notification).

### Why did U.S. Prudential Regulators require prior supervisory model approval for dealers using ISDA SIMM, while EU EMIR Phase 6 funds did not until EMIR 3.0?

Under U.S. Prudential and CFTC rules, all swap dealers needed formal regulatory approval (via NFA / Fed / OCC) to run ISDA SIMM; EU EMIR 3.0 (2024) newly introduced an EBA Centralized Pro-Forma Model Validation regime to harmonize SIMM oversight across European counterparties.

### How does ISDA SIMM treat basis risk between SOFR and Term SOFR or Fed Funds within the same USD currency bucket?

Inside the USD IR bucket, SIMM does *not* assume 1.000 correlation between different sub-curves (OIS SOFR vs Fed Funds vs Prime vs Municipal): cross-sub-curve sensitivities multiply by a sub-curve correlation scalar phi_{ij} < 1.0 (e.g., 0.982–0.993), charging explicit basis margin on massive SOFR/FF basis books!

### What is a "SIMM Crowding / Multiplicative Add-On" imposed by regulators on bespoke risk factors?

When backtesting shows that a counterparty’s exotic portfolio (e.g., correlation swaps, dividend futures, or weather derivatives) contains material risk uncaptured by SIMM’s standard CRIF buckets, supervisors require trade-level Schedule fallbacks or SIMM Multiplicative / Fixed Add-Ons.


## Source URLs

- https://www.isda.org/2023/09/19/isda-simm-methodology/
- https://www.bis.org/bcbs/publ/d499.htm
- https://www.eba.europa.eu/activities/single-rulebook/regulatory-activities/market-infrastructures/emir-initial-margin-model-validation

## Citation

- APA: Arthur Pendleton, CPA, JD. (2026). ISDA SIMM v2.6 / v2.7 Methodology: Delta, Vega, Curvature Risk Weights, Concentration Thresholds & Add-Ons. FinLexicon Wiki. https://finlexicon.org/wiki/isda-simm-v2-6-delta-vega-curvature-risk-weights-correlations
- Canonical article: [ISDA SIMM v2.6 / v2.7 Methodology: Delta, Vega, Curvature Risk Weights, Concentration Thresholds & Add-Ons](https://finlexicon.org/wiki/isda-simm-v2-6-delta-vega-curvature-risk-weights-correlations)
- Site research index: [FinLexicon Wiki](https://finlexicon.org/llms-full.txt)
