What's New in ISDA SIMM Version 2.9+2512?

ISDA has published the ISDA SIMM® Methodology, version 2.9, dated 22 July 2026 — eleven days after version 2.8+2512 became the live margin model. It arrives with no effective date attached, which is the clue to what it is for: this is the industry's advance sight of the next methodology change, not a switch that is about to be flipped.

SIMM changes arrive in two different flavours, and they are easy to confuse. A calibration refreshes the numbers — risk weights, correlations, concentration thresholds — against a new window of market data. A methodology change alters the machinery itself: how currencies are grouped, how buckets are defined, how the formulas fit together. Calibrations happen on a fixed schedule. Methodology changes are rarer, require regulatory notification, and are the ones that move margin in ways a portfolio manager does not expect.

Version 2.9+2512 is the new 2.9 machinery applied to the same December 2025 market data currently running live under 2.8+2512. Pairing a new methodology with a dataset that is already in production is a useful thing to do: because both sides sit on identical data, every difference is the methodology and nothing else — no market noise, no recalibration mixed in. It shows the shape of the coming change on its own, well before it counts.

This exact combination is not a margin version, and ISDA has not given it an effective date. In practice it lets firms run their books through the new mechanics, find the exposures that behave differently, and be ready before the model actually switches. The last methodology change followed the same sequence: version 2.8 was published against the outgoing 2412 dataset ahead of 2.8 going live in production.

ISDA recalibrated SIMM once a year until 2024, always with an early-December go-live. Since 2025 the model has run on a semiannual cycle, and in practice the two go-lives have settled on mid-July and early December, each falling on a Saturday with the preceding Friday's close of business the first calculation date:

• Primary calibration — a full recalibration of every parameter against market data to 31 December, effective the following mid-July. Most recently 2.8+2512, effective 11 July 2026.

• Secondary calibration — a narrower refresh, focused on the main delta risk weights, against data to 30 June, effective the following early December. Most recently 2.8+2506, effective 6 December 2025.

The natural home for version 2.9 is the next secondary calibration — the methodology recalibrated to market data through 30 June 2026, which would put it in market as 2.9+2606 with an early-December 2026 effective date. ISDA has not confirmed either the version name or the date, so treat this as the shape of the cycle rather than a date to plan against. Under the SIMM Governance Framework, ISDA commits to at least 60 calendar days' notice of any effective date, so confirmation will come well ahead of the switch.

This is the single most important thing to take from a preview release. What you are looking at below is the structure of SIMM 2.9, not the parameters you will eventually margin on. Before go-live, ISDA runs the next calibration and restates the numbers against a fresher data window. The mechanics you see here carry forward. The levels do not.

How much do they move? The 2.8 cycle is the one clean precedent, because it followed the same preview-then-production sequence. Our comparison of the 2.8 preview against the version that actually went live six months later shows:

• Roughly 40% of all calibration parameters moved between the preview and production.

• Of those that moved, the median change was about 6%, with the top decile shifting 18% or more and a handful of parameters more than doubling.

• The risk classes the methodology step had left completely untouched moved anyway — Credit Qualifying restated 83% of its parameters, Equity 56%, Interest Rate 50% — because a recalibration reaches everything, not only the parts the methodology changed.

• But the structural changes survived intact. The new FX volatility grouping that version 2.8 introduced was carried through to production and refined, not reversed.

That is one cycle of evidence, not a law, but it points to a sensible working rule for reading any preview: trust the structure, discount the levels. Bucket definitions, currency groupings and the shape of the correlation framework are decisions, and decisions tend to stick. Individual risk weights and thresholds are outputs of a data window that has not closed yet.

One point specific to this release. The 2.8 methodology step was a light touch — it moved a few FX risk weights and a handful of cross-class correlations, leaving Interest Rate, Credit, Equity and Commodity entirely alone. Version 2.9 is a far heavier step: it restates two thirds of the Interest Rate parameters and rebuilds the FX volatility universe outright. So while Equity, Commodity and Credit Qualifying show as flat in the comparison below, that is not a reason to expect a quiet go-live. They are simply untouched by the methodology, and on last cycle's evidence they are unlikely to stay flat once the next calibration lands.

• 2.9: the SIMM methodology version.
• +2512: calibration based on market data through December 2025.

Developed by Cumulus9, this tool highlights the quantitative impact of SIMM parameter changes across all asset classes. Use the dropdown menu to explore results by risk type, and interact with the charts and tables to analyze how the new version affects your margin requirements.

Interest Rate absorbs most of the change. Every regular-volatility delta risk weight falls, with the belly and long end cut hardest (5y 61→53, 10y 60→50, 15y 58→51) and the front end easing more modestly (2w 107→99, 3m 90→83). Low-volatility JPY moves the other way at the very front (2w 15→19, 1m 18→22) and is flat beyond 3m, while the high-volatility curve is re-shaped rather than re-levelled — 2w jumps 167→270, 1m–3m fall (102→74, 79→64), and the 1y–10y section drifts a few points higher. The inflation weight eases to 49 and the vega weight to 0.18 (from 0.20), with the cross-currency basis swap spread weight (21) and historical volatility ratio (0.74) unchanged. Correlations soften almost everywhere on the curve, and the correlation between different currencies drops sharply from 0.35 to 0.29, handing multi-currency rates books a materially larger diversification credit; inflation versus yield moves the other way, strengthening to 0.46. The currency groups are re-cut, with CNY joining the regular-volatility group and HKD leaving it. The sting is in concentration: the high-volatility delta threshold collapses from 71m to 17m, so concentrated exposures in high-volatility currencies hit the concentration multiplier far sooner, while the remaining delta thresholds are unchanged and the vega thresholds loosen slightly.

FX carries the single largest structural change in the release. The group of high FX volatility currencies expands from nine currencies to thirty-seven — KRW, ZAR, PLN, TRY, HUF, AMD, KZT, UAH, VES, ZWG and a long tail of frontier and restricted currencies join the incumbents, while SCR drops out. With a far broader and less extreme population, the risk weights attached to that group fall dramatically: High/High from 31.7 to 9.6 and High/Regular from 31.7 to 11.5, with Regular/Regular easing to 7.2. Correlations rebalance to match. Where the calculation currency is itself high-volatility, High/Regular falls to 0.63 and Regular/Regular falls sharply to 0.80 (from 0.97); where the calculation currency is regular-volatility, the picture inverts, with High/High rising to 0.20 (from 0.03) and High/Regular to 0.31 (from 0.12) — the enlarged group now behaves as a correlated block rather than a set of idiosyncratic outliers. The concentration thresholds are untouched. The practical consequence is that many more currencies now sit inside the high-volatility group at a much lower weight, and the higher cross-group correlations claw back part of the diversification benefit the old narrow definition granted. This is the change most worth testing against your book: if you run emerging-market or frontier FX, your exposures have changed category.

Credit Non-Qualifying sees a single, contained change: the Investment Grade delta risk weight rises from 210 to 220, roughly 5%. High Yield, correlations and the concentration framework are all unchanged. Credit Qualifying, Equity and Commodity are unchanged in full — same delta and vega weights, same correlation matrices, same concentration thresholds — which is what you would expect from a methodology step that does not reach them, and which says nothing about where they land at the next calibration. Cross-risk-class correlations move only where Interest Rate is involved: IR/Commodity strengthens 0.34→0.37 and IR/Credit Qualifying 0.09→0.10, while IR/FX weakens 0.15→0.12, IR/Equity 0.11→0.10 and IR/Credit Non-Qualifying 0.07→0.06.

Overall, version 2.9 is a targeted rework of the two risk classes where currency grouping drives the numbers. Rates books get lower per-position margin and a materially better cross-currency diversification credit, offset by a far sharper concentration charge on high-volatility currencies. FX books face a much wider high-volatility universe at much lower weights, with the net direction depending on which currencies a portfolio actually holds. The time to run these structural changes against your book is now, while they can be seen in isolation — at go-live they will arrive bundled with a restatement of the numbers.

Parameter-level impact only takes you so far. What matters is what version 2.9 does to your portfolio, and that depends entirely on where your risk sits — which currencies, which tenors, how concentrated. Clients calculating margin through the Cumulus9 API can measure that today: version 2.9 is available on demand alongside every previous release, so the same book can be priced under both methodologies and the difference read straight off.

Set simm_metrics in the portfolio payload to { "version": "2_9_2512", "holding_period": 10 } and the calculation runs on the new methodology. Omit it, or leave it at 2_8_2512, and you get the live production version — the default is unchanged, so an impact run costs nothing more than a second call and touches nothing in your normal flow. Full payload details are in the portfolios endpoint documentation.

As a licensed ISDA SIMM vendor, Cumulus9 is here to support you with advanced risk analytics and tailored insights into the latest SIMM methodology and calibration updates. Our expertise ensures you have the tools and information necessary to respond proactively to margin changes. Contact us to discover how our solutions can help you optimize risk management in today's shifting market landscape.