Initial Margin for Crypto Risk in Uncleared Markets (FEDS 2026-009)

Why Crypto Needs Its Own Risk Class in ISDA SIMM

The Federal Reserve FEDS 2026-009 paper by Anna Amirdjanova, David Lynch, and Anni Zheng addresses a growing gap in financial risk management: how to properly calculate initial margin for derivatives that are sensitive to cryptocurrency risk factors in uncleared markets. As the crypto market has surged to approximately $4 trillion in total capitalization — with CoinMarketCap tracking over 18,815 active cryptocurrencies — the existing ISDA Standardized Initial Margin Model (SIMM) framework lacks any dedicated treatment for crypto risk.

This isn’t merely an academic exercise. SIMM is the dominant model used for initial margin computation in the uncleared derivatives market, accounting for over 90% of all initial margin collected and posted. Total initial margin in uncleared markets has stabilized around $431 billion in both 2023 and 2024. As institutional exposure to crypto derivatives grows, the absence of a crypto-specific risk class within SIMM creates the potential for systematic under-margining — a concern for both individual counterparties and broader financial system stability.

The paper’s central recommendation is both clear and consequential: cryptocurrencies should be classified into a distinct risk class within SIMM, separate from commodities, FX, or any other existing category. This new class should be split into two buckets — floating (unpegged) cryptocurrencies and pegged (stablecoin) cryptocurrencies — with calibrated risk weights and correlations specific to each bucket’s behavior.

The Uncleared Margin Rule and SIMM Framework Explained

To understand why this paper matters, it helps to grasp the regulatory architecture it operates within. The Uncleared Margin Rule (UMR), implemented globally in phases since 2016, requires market participants to exchange initial margin on non-centrally-cleared derivatives. The goal is straightforward: reduce systemic risk by ensuring that counterparties hold sufficient collateral to cover potential losses during the close-out period if one party defaults.

ISDA SIMM was developed as the industry-standard model for these calculations. It’s a sensitivities-based approach designed to target the 99th percentile of potential losses over a 10-day margin period of risk (MPOR). The model organizes financial instruments into risk classes — currently including interest rates, FX, equities, commodities, and credit — with each class having its own set of risk weights, correlations, and calibration periods.

The key challenge the paper addresses is that when crypto-sensitive derivatives are traded in uncleared markets, there’s no standardized, validated method to compute their initial margin under SIMM. Some market participants have attempted to shoehorn crypto into the commodities risk class, but as the paper demonstrates, this approach fundamentally underestimates the risk involved.

Floating vs. Pegged: The Two-Bucket Crypto Classification

One of the paper’s most important contributions is the formal distinction between floating and pegged cryptocurrencies as fundamentally different types of risk factors. This distinction has profound implications for margin calculations.

Floating (Unpegged) Cryptocurrencies

The floating bucket includes major cryptocurrencies that derive their value from market dynamics: Bitcoin (BTC), Ethereum (ETH), Binance Coin (BNB), Cardano (ADA), Dogecoin (DOGE), and Ripple (XRP). These assets exhibit extreme volatility relative to traditional asset classes. Bitcoin alone accounts for approximately 58.5% of total crypto market capitalization, with Ethereum contributing another 12%.

The paper finds that floating cryptocurrencies have a high intra-bucket correlation of approximately 73% (using the Kendall-to-Pearson conversion method consistent with SIMM practice). This means these assets tend to move together during stress periods — a critical consideration for portfolio margining.

Pegged (Stablecoin) Cryptocurrencies

The pegged bucket includes stablecoins designed to maintain a 1:1 peg with the US dollar: USDT, USDC, DAI, TUSD, USDP, and GUSD. Despite their design intent, these instruments carry their own set of risks — de-pegging events, liquidity crises, and issuer-specific operational risks.

Critically, pegged cryptocurrencies show a much lower intra-bucket correlation of only 14-20%, depending on the stress-period selection method. This dramatically lower correlation reflects the idiosyncratic nature of stablecoin risks — each stablecoin’s de-peg risk depends on its specific backing mechanism, issuer, and regulatory environment, rather than on common market factors.

The inter-bucket correlation between floating and pegged crypto is essentially zero. Sample estimates cluster around -2% to -4%, and the authors recommend setting this to zero in practice — meaning the two buckets provide virtually no margin offset against each other.

Data and Methodology: Twelve Cryptocurrencies Under the Microscope

The paper’s empirical analysis draws on price data for twelve major cryptocurrencies — six floating and six pegged — sourced primarily from CoinGecko, CoinMarketCap, and BitBo. The sample period varies by instrument, with Bitcoin data extending back to January 2013 and all series ending on July 17, 2025.

These twelve instruments were chosen because they collectively represent approximately 85% of total crypto market capitalization. Bitcoin’s market cap in the sample stands at an impressive $2.36 trillion, with Ethereum at $426 billion. Among stablecoins, USDT dominates with a market cap exceeding $160 billion and daily trading volume also above $160 billion.

The calibration methodology follows SIMM’s established approach: combine a recent 3-year observation period with an additional stress period. The stress period is identified by selecting the top 10% most volatile disjoint quarters from the available history. For the SIMM 2.7+2412 calibration referenced in the paper, this meant selecting the six most volatile quarters from January 2, 2008 through December 31, 2024.

For correlation estimation, the paper adopts Kendall’s tau — a nonparametric rank correlation measure — converted to Pearson-equivalent via the formula sin(0.5π·τ). This is the same approach used in SIMM’s existing calibrations and is chosen specifically for its robustness to outliers. The paper includes simulation evidence showing that when two large outliers are introduced into sample data, raw Pearson correlation drops from 0.65 to 0.32, while the Kendall-to-Pearson method remains stable at approximately 0.63.

Stress-Period Selection: Why the Greedy Algorithm Wins

A particularly nuanced aspect of the paper is its comparison of two approaches for selecting the stress calibration period for crypto buckets.

Approach 1: Single Pseudo-Index (BGCI)

The first approach uses the Bloomberg Galaxy Crypto Index (BGCI) as a pseudo-index for the entire crypto class, selecting stress periods based on the volatility of this index. While simpler, this approach has a significant limitation: BGCI is dominated by floating crypto (primarily BTC and ETH), making it a poor representation of stablecoin dynamics.

Approach 2: Bucket-Specific Greedy Algorithm

The second approach applies the ISDA “greedy” algorithm independently to each bucket’s calibration instruments. The greedy algorithm works by computing an R-score for each candidate period: R(i) = Σ_k[vol(k,i)/MaxVol(k)], where vol(k,i) is the average daily volatility for instrument k during period i, and MaxVol(k) is the maximum such volatility across all candidate periods. The algorithm then selects the disjoint quarters with the highest R-scores.

This bucket-specific approach identifies different stress periods for floating and pegged buckets. For floating crypto, top stress quarters include periods like November 2017 – February 2018 (the first major Bitcoin bubble). For pegged crypto, the stress period selection identifies quarters such as March – June 2020 (coinciding with the COVID-19 market shock that tested several stablecoin pegs).

The paper favors the greedy algorithm approach, as it produces more accurate and bucket-appropriate calibration — particularly important for ensuring pegged crypto stress periods reflect actual stablecoin-specific stress rather than floating crypto volatility spikes.

Delta Risk Weights: Commodity Class vs. Dedicated Crypto Class

Here’s where the numbers speak most powerfully. The paper presents a direct comparison of delta risk weights under different classification scenarios:

The implications are stark: placing floating crypto in its own risk class more than doubles the delta risk weight — from 58 to approximately 132. This means that if a dealer uses the commodity classification, they would hold roughly half the margin that the crypto-specific classification requires. In a market where single-day moves of 10-20% are not uncommon for major cryptocurrencies, this under-margining represents a genuine counterparty risk concern.

For pegged crypto, the difference is less dramatic but still meaningful: a risk weight of 1 under commodity classification versus 2 under the bucket-specific greedy approach. While small in absolute terms, this difference matters for large stablecoin derivative positions where notional amounts can be substantial.

Delta risk weights in SIMM are constructed as the median across instruments of the maximum of the absolute 1st and 99th percentiles of overlapping 10-day relative returns over the calibration period. The much higher risk weight for floating crypto in a dedicated risk class reflects the selection of crypto-appropriate stress periods that capture the full magnitude of crypto-specific volatility events.

Correlation Analysis: Crypto’s Low Cross-Asset Linkage

One of the paper’s most practically relevant findings concerns the correlation between cryptocurrency risk factors and traditional asset classes. Using the Kendall-to-Pearson methodology, the paper computes cross-risk-class correlations that are uniformly low:

Using USDT as Crypto Representative

• FX: ~12%
• Interest Rates: ~11%
• Equities: ~6%
• Commodities: ~5%
• Credit Qualifying: ~4%
• Credit Non-Qualifying: ~3%

Using BGCI as Crypto Representative

• Interest Rates: ~13%
• Credit Qualifying: ~11%
• Commodities: ~9%
• Equities: ~2%
• FX: ~2%
• Credit Non-Qualifying: ~2%

These low correlations have a critical practical implication: firms cannot achieve significant margin reduction by cross-hedging crypto exposure with positions in traditional asset classes. Under SIMM’s aggregation formula, low cross-class correlations mean the crypto risk component adds nearly independently to total initial margin requirements. This challenges any strategy built on the assumption that crypto provides meaningful diversification benefits within a derivative portfolio context.

The variation between USDT-based and BGCI-based correlation estimates also highlights the importance of choosing the right representative index for the crypto risk class in cross-asset calculations — another area where the paper provides concrete guidance.

Policy Implications for Derivatives Markets

The paper’s recommendations carry significant weight for the future of crypto derivatives regulation and risk management. If adopted by ISDA and endorsed by regulators, the framework would:

• Increase initial margin requirements for crypto-sensitive uncleared derivatives significantly — potentially doubling margins for floating crypto exposure compared to current commodity-class treatment.
• Create a standardized framework that eliminates the current inconsistency where different market participants may classify crypto risk differently within SIMM.
• Separate stablecoin risk treatment from floating crypto, recognizing that these instruments face fundamentally different risk profiles despite both being “crypto.”
• Limit cross-asset margin offsets by establishing the empirical reality that crypto has very low correlation with traditional risk classes.
• Provide a calibration methodology that can be updated as the crypto market matures and more data becomes available — including the potential addition of new buckets as new categories of digital assets emerge.

For financial institutions and compliance teams, the timing is important. As more institutional players enter crypto derivatives markets, the gap between actual risk and margined risk becomes a regulatory and operational priority. The paper provides the quantitative foundation for closing this gap.

What This Means for Market Participants and Regulators

The practical implications of adopting this framework extend across the derivatives ecosystem. For dealers and buy-side firms with crypto derivative exposure, higher initial margin requirements would increase the cost of maintaining uncleared positions — potentially shifting more activity toward centrally cleared venues where they exist.

For risk managers, the paper provides a validated methodology for crypto risk quantification within the SIMM framework. The bucket-specific stress-period selection via the greedy algorithm is particularly useful, as it avoids the trap of using a one-size-fits-all calibration that may not capture the unique stress dynamics of different crypto sub-categories.

For regulators, the framework offers a systematic approach to supervising crypto margin adequacy. The paper’s finding that commodity-class placement roughly halves the appropriate risk weight for floating crypto quantifies the under-margining concern that has been discussed qualitatively in regulatory circles.

For stablecoin issuers and DeFi protocols involved in institutional derivatives, the separate treatment of pegged crypto risk factors validates the view that stablecoins are a distinct financial instrument — but also confirms that they carry their own idiosyncratic risks that cannot be ignored in margin calculations.

The research also points to areas requiring further work: how to treat crypto options (the paper focuses on delta sensitivities), how to incorporate DeFi-specific risks, and how the framework should evolve as tokenized traditional assets blur the line between crypto and conventional risk factors. With the total crypto market now rivaling the GDP of major economies, getting these margin calculations right is no longer optional — it’s a cornerstone of financial stability.