M.2Reference Portfolio

The analysis applies to a representative book of non-centrally cleared OTC derivative transactions composed of seven product types, each at $10,000,000 USD-equivalent notional — four rates and FX products carried from Phase 1, and three Bitcoin reference products added under Phase 2 (WS1):

• USD Interest Rate Swap, 5-year tenor. Vanilla SOFR fixed-for-floating, daily compounded SOFR vs annual fixed coupon, ACT/360 day count.
• USD Interest Rate Swap, 10-year tenor. Vanilla SOFR fixed-for-floating, same conventions as the 5Y leg.
• EUR/USD FX Forward, 1-year tenor. Single-currency-pair forward purchase of EUR vs USD at forward points as of the Calibration Date.
• USD/EUR Cross-Currency Swap, 5-year tenor. Floating-for-floating USD-SOFR vs EUR-ESTR + basis spread, quarterly resets.
• BTC/USD Forward, 1-year tenor. Bilateral forward purchase of Bitcoin against US dollars at a pre-agreed price. Single bullet exchange at maturity; the clean reference case against which the other two crypto products are measured.
• BTC Perpetual Swap. Swap on the Bitcoin price with no contractual maturity. Economics are carried by a funding rate exchanged three times daily between longs and shorts, calibrated to the 30-day trailing average of realised settlements (6.658% annualised at the Calibration Date, 90 observations).
• Bitcoin Total Return Swap, 1-year tenor. One party receives the total return on Bitcoin and pays a financing leg of SOFR plus a spread (425 bp, the derived dealer floor of 8.418% — §M.15.7). Capital treatment is identical to the dated forward; the financing leg is a P&L rather than a capital effect.

The portfolio spans short-dated single-currency through long-dated cross-currency, intentionally covering the spectrum of XVA capital intensities relevant to a typical dealer non-cleared derivatives book. Calibrated par rates, forward points, and basis spreads are locked against the Calibration Date market snapshot (7/31/2026) and documented in §M.13.

Trade parameters (locked at the Calibration Date, 7/31/2026)

Engine base values per trade (Regime A baseline, per $10M notional — headline displays per $1B, ×100)

Reading the crypto rows. Three features look anomalous against the rates and FX products and are not errors.

• SIMM IM exceeds notional. At a delta risk weight of 132 the margin on a $10M linear position is $13.2M. The weight is calibrated to the median of the maximum absolute 1st/99th percentile of 10-day overlapping relative returns over three recent years plus the November 2017–November 2018 stress year — a window containing 10-day gains above 100%. The figure is prospective in any case; see §M.6.
• SA-CCR EAD is identical across all three. SCO60.98 gives the crypto add-on the FX asset-class structure, which carries no supervisory duration, so tenor does not enter the calculation. The dated forward, the perpetual and the TRS therefore share one EAD; see §M.7.
• The perpetual's BA-CVA, MVA and KVA are an order of magnitude smaller. Those four legs scale with effective maturity, and a perpetual has none. The values shown use the default convention — the margin period of risk itself, M_eff = 10/250 = 0.04 years, treating the position as spot risk held over the close-out window. The tool exposes 1-year rolling and a 0.5-year floor as alternatives; selecting either moves these four figures and leaves IM, EAD and PFE unchanged.

The forward and the TRS are identical on every capital measure. What separates them is the financing leg, which is a P&L rather than a capital effect and is set out per product in the SA-CCR derivation drawer.

M.3Counterparty Contexts

The analysis addresses two counterparty contexts, each governed by a standard ISDA Master Agreement and CSA:

D2D — Dealer-to-Dealer

Bilateral institutional context between two derivatives dealers. Standard ISDA + CSA terms assumed: daily margin, zero threshold, zero minimum transfer amount (MTA), bilateral close-out, eligible currencies USD and EUR. SIMM applies to both parties bilaterally.

D2BS — Dealer-to-Buy-Side

Bilateral context between a derivatives dealer and a buy-side counterparty falling under Uncleared Margin Rules (UMR) phase 6 (firms with AANA > €8B). Standard ISDA + CSA terms assumed: daily margin, zero threshold, $250,000 MTA, bilateral SIMM IM exchange.Buy-side context introduces additional operational complexity (estimated additional IM management, documentation, and tri-party custodian costs typical of Phase 6 UMR compliance) reflected in the analysis as an indicative 12% uplift applied uniformly to metric values. (Please note: this reflects a calibration assumption rather than an output of the underlying models.)

M.4Collateral Regimes

The analysis compares four collateral regimes, each defined by the set of permitted collateral types under the bilateral CSA:

• Regime A — Baseline. Cash + US Treasury securities only. Treated as the regulatory baseline (full netting eligibility, HQLA Level 1, margined leverage treatment per §M.12).
• Regime B. Cash + USTs + USDC/USDT private-issuer stablecoins. Stablecoins generally do not satisfy UCC Article 8/9 perfection or bankruptcy safe-harbor prongs without bespoke documentation, resulting in non-nettable treatment for derivative collateral purposes.
• Regime C. Cash + USTs + tokenized money market fund shares (TMMF), with BUIDL as representative. TMMF qualifies for partial recognition via look-through to the underlying fund interest, provided the fund interest is itself eligible.
• Regime D — Focus. Cash + USTs + USDM1. USDM1 satisfies both netting prongs via Marshall Islands legislation incorporating UCC Articles 8/9 and explicit bankruptcy safe-harbor treatment, plus potential IM eligibility under UMR.

M.5Eligibility Frameworks

UCC Article 8/9 Perfection

Under Article 8 of the US Uniform Commercial Code, a "securities entitlement" is created when a financial asset is credited to a securities account at a securities intermediary. Article 9 governs perfection of security interests in those entitlements. For collateral to be effective in a bilateral derivative netting context, the secured party (collateral taker) must have a perfected security interest under the law of the applicable jurisdiction. USDM1, issued under Marshall Islands legislation that incorporates UCC Articles 8 and 9 by reference, satisfies both elements when held at a qualifying intermediary.

Bankruptcy Safe-Harbor

The US Bankruptcy Code includes safe-harbor provisions (primarily under §§555, 559, 560, 561) that protect close-out netting and collateral liquidation rights for qualifying financial contracts, notwithstanding the automatic stay or anti-deprivation principles that ordinarily apply in bankruptcy.

Standard ISDA Master + CSA close-out is preserved under these safe-harbors where the collateral is "eligible" under the relevant definition. USDM1 receives express safe-harbor treatment by virtue of its classification as a sovereign-issued securities entitlement.

USDM1’s indenture and security documents are explicit in their governing-law architecture. USDM1 is a New York-law instrument with an explicit customary waiver of sovereign immunity.

This structure provides the secured party dual recourse — a claim against the issuer before New York courts for guaranteed par redemption, and enforcement of the perfected first-priority Treasury pledge through the instrument’s US trustee, custodian and collateral agent (Surus Trust Company, a US-chartered trust company).

These dynamics exist alongside the federal Bankruptcy Code safe-harbor protections.

Although the RMI maintains its own netting legislation, USDM1’s structure places netting and enforcement under a US framework rather than a foreign-issuer framework.

Full collateralization by short-dated US Treasuries held by a US trust company and qualified custodian, with a perfected first-priority security interest for the benefit of holders, further removes residual foreign-issuer risk.

Collateral may be subject to customary limited liens in favor of a custodian. The Fed/OCC/FDIC interagency FAQ on capital treatment of tokenized securities (5 March 2026) has a financial-collateral condition that expressly accommodates the prior security interest of a custodial agent (FAQ fn. 4). Accordingly, customary custodian liens do not defeat financial-collateral qualification.

Operationally, USDM1 is accommodated under standard ISDA CSA eligible-collateral schedules (and GMRA/GMSLA annexes for repo and securities-lending use) by schedule election; no master-agreement amendment is required.

CFTC §23.156 Eligibility

CFTC §23.156 establishes eligible collateral types for initial margin posting under the Uncleared Margin Rules. The standard list includes cash, US Treasury securities, certain government and agency securities, and specified money market fund shares. Tokenized sovereign instruments such as USDM1 provide structural equivalence to the enumerated eligible collateral types and may have a path to explicit relief.

Basel 3.1 Output Floor

The Basel 3.1 output floor (BCBS d424) constrains aggregate risk-weighted assets computed under internal models to no less than 72.5% of RWA computed under the standardized approaches. The BCBS timetable phases it to 1 January 2028 (GHOS deferral of 27 March 2020); the transitional running to 2030 is the EU implementation under CRR3 Art. 465. This analysis applies standardized approaches throughout (SA-CCR for EAD, BA-CVA for CVA capital, standardized supervisory factors), so the output floor is not a binding constraint on the quantified metrics. Banks subject to the output floor that compute these metrics under internal models would observe equal or smaller capital benefits than those reported here, reflecting the standardized-approach floor.

M.6SIMM IM Methodology

Initial Margin is computed under ISDA SIMM v2.8, calibration +2512 — the version in force at the Calibration Date, effective 11 July 2026 and superseding v2.8+2506 — the Standard Initial Margin Model. SIMM is the industry-standard methodology adopted under the global Uncleared Margin Rules (UMR), including US CFTC and prudential regulator margin rules, for the calculation of regulatory initial margin on non-centrally cleared derivative transactions between covered counterparties.

Framework

SIMM applies a sensitivity-based aggregation at the asset-class level (interest rate, foreign exchange, equity, commodity, credit qualifying, credit non-qualifying). For each asset class, the IM contribution is:

K_assetclass = K_delta + K_vega + K_curvature + K_baseCorr

Each K_x is computed as a weighted sum of risk sensitivities (s_i) and ISDA-published risk weights (RW_i), aggregated through the prescribed correlation matrices (intra-bucket and inter-bucket; the cross-class and cross-currency parameters used here are ISDA SIMM v2.8+2512 §K ¶88 and §D.2 ¶37 respectively, tabulated below).

Total SIMM IM aggregates the risk-class contributions under the prescribed inter-class correlation matrix ψ:

SIMM IM = √( Σ_b K_b² + 2 Σ_{b<c} ψ_bc K_b K_c )

Critical property: regime invariance

SIMM IM is a function exclusively of (a) the trade's risk sensitivities, (b) the ISDA-published risk weights and correlations, and (c) the SIMM aggregation methodology. None of these inputs depend on the collateral posted against the trade. The SIMM IM dollar amount for a given trade and counterparty pair is therefore invariant across the four collateral regimes (A, B, C, D) examined in this analysis.

The regime variation in this analysis lives in the CRM-driven metrics (SA-CCR EAD, CVA, FVA, KVA, PFE), where collateral type changes the recognized risk mitigation under Basel CRE22.40–22.65. SIMM-side regime variation is operational, not numerical: USDC/USDT cannot be posted as SIMM IM under §23.156, so a bank holding USDC/USDT must additionally post cash or US Treasuries against the same SIMM amount, producing dead capital on the balance sheet. USDM1 has a path to relief whereby it could enter the eligible IM universe and support direct posting against this SIMM amount.

Reference portfolio calibration

For each of the four rates and FX products in §M.2 (the crypto set is covered separately in §M.6b below):

• USD IRS 5Y at $10M: K_IR_delta = |IR Delta × RW_USD_5Y| = |$4,552 × 61 bp/bp| = $277,672. Validated against par-swap pricing at Calibration Date.
• USD IRS 10Y at $10M: K_IR_delta ≈ |$8,127 × 60 bp/bp| = $487,620 (indicative; IR Delta confirmation pending).
• EUR/USD FX Forward 1Y at $10M: K_FX_delta ≈ RW × FX delta aggregated under SIMM §I.1 ¶69 (regular/regular FX volatility group); calibrated magnitude ~$80K (indicative).
• USD/EUR XCCY 5Y at $10M: K_XCCY = sqrt(K_USD_IR² + K_EUR_IR² + 2·ρ·K_USD·K_EUR + K_FX² + cross-terms) under SIMM §D.2 ¶37 (γ=0.35 cross-currency) and §K ¶88 (IR/FX 0.15). Note that the IR pair alone returns ~$466K at γ=0.35, above the ~$455K carried in the engine base values (§M.2): the $455K is a calibrated indicative magnitude, not the output of the operands shown, and reproducing it from this form would require γ ≈ 0.287. The aggregation form is stated because it is the correct construction; the level is asserted, per §M.6 Calibration status.

Vanilla USD SOFR fixed-for-floating IRS and FX forwards are linear products with no optionality; SIMM vega and curvature components are zero for these trades. XCCY 5Y is also linear (floating-for-floating); no optionality contribution.

Risk weights applied

M.6b · Crypto risk class — prospective construction

Corresponds to §M.15.3 of the methodology document, which carries the fuller treatment.

Risk weights follow FEDS 2026-009 (Amirdjanova, Lynch and Zheng, 11 February 2026), which calibrates a crypto risk class on the SIMM methodology's own terms. The Federal Reserve notice states that the analysis does not indicate concurrence by the Board.

The three reference products are linear in the Bitcoin price, so vega and curvature are zero and only the delta component contributes. Sensitivity is defined per 1% relative price move (ISDA US Patent 10,515,410), so for a $10M position s = $100,000 and:

K_crypto = s × RW = $100,000 × 132 = $13,200,000  ≡  notional × RW/100

Why the separate class matters. Placing crypto in the existing commodity risk class instead — the nearest available home under published SIMM — gives the floating bucket a delta risk weight of 58, less than half of 132. That gap is the central finding of the paper: the commodity class is not a conservative proxy for cryptoasset price risk, it is a materially lighter one. For scale on identical units, the same table gives European Power 64 and Coal 48.

All three crypto products share one bucket and one risk factor, so intra- and inter-bucket correlations do not bind on the reference portfolio as configured; they are carried because they bind as soon as a second cryptoasset or a pegged position enters the netting set.

Calibration status

IRS 5Y SIMM IM is validated against par-swap pricing. Crypto IM is prospective — it depends on a risk class existing in published SIMM, not on a market-data confirmation.

M.6.1Incremental Initial Margin — the Margin Tools construction

The incremental cost of adding a trade T to an existing netting set S follows directly from the aggregation of §M.6:

IM(S ∪ T) = √( K_S² + K_T² + 2·ψ·K_S·K_T )    incremental = IM(S ∪ T) − K_S

with ψ the cross-risk-class correlation between the incoming trade's class and the existing set's. The offset the calculator reports is 1 − incremental / K_T.

Three properties, all visible in the tool

• Zero against an empty set. There is nothing to diversify against, and the incremental cost of the first trade is its standalone margin.
• Where ψ < 1, it rises as the existing set grows relative to the incoming trade, tending to 1 − ψ in the limit where the set dominates. Where ψ = 1 it stays at zero throughout.
• Invariant to the collateral posted, because every input to §M.6 is — the standalone, incremental and post-trade figures are identical across the four collateral regimes.

This is not the book-unification quantity

The methodology document tabulates a different measure at its §M.17.4: the margin released by unifying a whole book, which is single-peaked in sleeve share (about 24.9% near a 1.9% sleeve, 18.6% at 5%, 1.6% at 50%). The quantity on this tab is the marginal cost of a single trade against an existing set, which is monotonic in K_S/K_T. They are different measurements of the same underlying diversification and should not be read against each other.

Stated assumption

The calculator takes the existing set as a single aggregate contribution in the interest-rate class. That is a modelling convention, not a measurement of the reference book: on the §M.17.1 book the rates sleeve dominates by notional, but the crypto sleeve carries the larger share of the initial margin, as the Cross-Asset tab states. The consequence is disclosed rather than buried — a crypto or FX trade added to this set is treated as cross-class and receives an offset, where a set modelled on its margin composition would make that same trade within-class and give it none. Where the incoming trade is itself interest-rate — the 5Y and 10Y swaps, and the cross-currency swap, which is attributed wholly to that class — the within-class convention applies: margin is summed rather than aggregated under the intra-class correlations, so ψ = 1, the reported offset is zero and the incremental cost is the full standalone margin. SIMM would recover a real benefit there, between the 5Y and 10Y buckets in particular, so that zero is a floor and not an estimate of the true offset. Interest-rate trades therefore show no diversification on this tab by construction, and only the cross-class pairings report a benefit — those being the ones the §M.6 calibration supports.

Where collateral does enter

It enters at eligibility, and nowhere else. Collateral that cannot satisfy the initial-margin requirement leaves that requirement to be met again in eligible form — the encumbrance the tool reports on its final row. The advantage claimed for USDM1 on this tab is eligibility, not a margin discount.

M.7SA-CCR EAD Methodology

Exposure at Default is computed under the Basel Standardized Approach to Counterparty Credit Risk per CRE52, the framework applicable to all non-internal-models banks and the binding constraint under the Basel 3.1 output floor for internal-models banks (BCBS d424, 72.5%; phased to 1 January 2028 under the BCBS timetable, to 2030 under EU CRR3 Art. 465).

Framework

SA-CCR computes EAD per netting set as:

EAD = α × (RC + PFE_addon)

where:

• α = 1.4 (CRE52.1, regulatory scalar applied uniformly to all netting sets)
• RC = max(V − C, 0) is the replacement cost — the net mark-to-market of the netting set minus net collateral posted. CRE52.10 gives this unmargined form; the reference book is daily-margined, where CRE52.18 governs and adds the threshold/MTA terms. For at-par trades with daily margin and no threshold, both reduce to RC = 0.
• PFE = multiplier × Σ AddOn_HS aggregated across hedging sets (CRE52.20). For at-par trades with no excess collateral, multiplier = 1.

The hedging set add-on for each asset class is:

AddOn = SF × ED × MF

• SF = supervisory factor by asset class (CRE52.72 Table 2): IR 0.50%, FX 4.00%.
• ED = effective notional. For IR derivatives: ED = δ × d × N (adjusted notional per CRE52.34; effective-notional composition per CRE52.30(1)), where δ is the supervisory delta (±1 for vanilla swaps), d is the supervisory duration, and N is the trade notional. For FX derivatives: ED = trade notional in trading currency (CRE52.35).
• MF = maturity factor (CRE52.52). For margined netting sets: MF = (3/2) × √(MPOR/250). At MPOR = 10 business days (CRE52.50 floor), MF = 0.300.

Supervisory duration for IR derivatives:

d = (e^(−0.05 × S) − e^(−0.05 × E)) / 0.05

where S is the start date and E is the end date in years. For IRS 5Y at par: d = (1 − e^(−0.25))/0.05 = 4.424.

Reference portfolio EAD (D2D, Regime A baseline, daily margin)

XCCY 5Y aggregates three hedging sets (USD IR, EUR IR, FX) per CRE52.45. IR hedging-set add-ons are summed, not correlated — CRE52.54 states supervisory correlations do not apply to interest-rate or FX derivatives, and CRE52.57(7) aggregates hedging-set add-ons by simple summation, superseding ρ = 0.5; FX is a separate asset class and adds directly.

CRM treatment by regime (CRE22.40–22.65 Comprehensive Approach)

The Comprehensive Approach to Credit Risk Mitigation under CRE22.40–22.65 recognizes eligible collateral with prescribed supervisory haircuts. Cash, US Treasuries, and qualifying sovereign debt are explicitly recognized; private-issuer stablecoins are not.

M.7c · Regime multipliers on the funding and exposure legs — status

Corresponds to §M.7.6 of the methodology document.

The table above derives the multiplier applied to SA-CCR EAD from the CRE52.48/52.52 maturity-factor transition (MF 0.300 → 1.000), and BA-CVA follows it one-for-one because SCVA is linear in EAD. Two further metrics also carry a collateral-regime multiplier, and their values are not derived from that transition:

USDM1 CRM interpretation

USDM1 is not specifically enumerated in CRE22.34/22.45 (the collateral schedules under the Comprehensive Approach to CRM). The defensible interpretation advanced in this analysis treats USDM1 as a Marshall Islands sovereign instrument under CRE22.34(3)(a)(i) (sovereign debt CRM treatment), conditional on:

Interagency confirmation of technology neutrality (Fed/OCC/FDIC FAQ, March 5, 2026).The three US prudential agencies jointly issued FAQs confirming that the capital rule is technology-neutral: an eligible tokenized security — one that, under applicable law, confers legal rights identical to those of the non-tokenized form — receives the same regulatory capital treatment as the non-tokenized form, and qualifies as financial collateral under the capital rule where the holder maintains a perfected, first-priority security interest.For this analysis:(a) the onchain form factor is, by itself, no longer a basis for denying CRM recognition — the technology objection is removed.(b) the legal threshold is tied to whether the rights conferred are identical in substance to a directly held sovereign claim - this is where the §M.5 perfection and safe-harbor conditions sit.The FAQ is drafted by reference to the issuance and transfer mechanism (distributed ledger technology). Accordingly, natively issued onchain instruments fall within its scope.

• UCC Article 8 securities entitlement is established under the Marshall Islands enabling legislation, and is recognized under NY-governed CSAs (M.5 conditional);
• Bankruptcy safe-harbor eligibility is established under US Bankruptcy Code §362(b)(17) — the stay exemption for a swap participant exercising rights under a security agreement or other credit enhancement related to a swap agreement — together with §560 and §561, for financial contracts secured by USDM1 collateral (M.5 conditional);
• USDM1 has a path to being named in eligible-collateral schedule for non-cleared OTC margin (M.5 pending).

Under these conditions, USDM1 receives Treasury-equivalent CRM treatment under CRE22.34(3)(a)(i) with the haircut of underlying US Treasuries (0.5–2% by maturity bucket).Certain user-driven tool configurations assume all three conditions are met.Additional user-driven tool configurations also provide data on alternative states, quantifying the impact if these conditions are not collectively met.

M.7b · Crypto asset class — SCO60.98

Corresponds to §M.15.1 (exposure), §M.15.2 (CVA) and §M.15.6 (collateral eligibility) of the methodology document. The perpetual's conventions are set out at §M.15.4.

The Basel cryptoasset standard SCO60 has been in force since 1 January 2026. SCO60.98 extends SA-CCR to cryptoasset derivatives by placing them in line with the FX asset class: the adjusted notional is the crypto notional expressed in domestic fiat, and there is no supervisory duration. Supervisory delta and the maturity factor follow the general rules.

For each of the three reference products at $10M notional, with the margined maturity factor MF = 3/2 × √(10/250) = 0.300:

AddOn = SF × ED × MF = 0.32 × $10,000,000 × 0.300 = $960,000
EAD = α × (RC + AddOn) = 1.4 × (0 + $960,000) = $1,344,000

RC is zero at par with daily margin. Because the FX structure carries no supervisory duration, this EAD is identical for the dated forward, the perpetual and the TRS — tenor does not enter the add-on at all. Note that the maturity factor keeps the Basel business-day convention (10/250); the calendar-day clock described in §M.11 applies to volatility scaling, which is a market-convention question rather than a supervisory-formula one.

Calibration status

SA-CCR formula and supervisory factor application is exact per CRE52.1–52.52, and per SCO60.98 for the crypto asset class. EAD magnitudes for the seven reference products are computed from standard supervisory inputs.

M.7.5Settlement Model & Margin Period of Risk (Forward-Looking Scenario)

Important. This methodology section documents a forward-looking scenario analysis exposed in the tool configurations via the "T+0 · 1d Settlement" toggle.The capital reduction quantified in the "Onchain T+0 · 1d MPOR" state is not currently recognized under Basel CRE52.50 for non-centrally cleared OTC derivative netting sets. The scenario is presented for analytical purposes only and is conditional on a future supervisory grant of margin-period-of-risk relief.

Margin Period of Risk under Basel CRE52.52

For margined netting sets, the SA-CCR maturity factor adjustment to the PFE add-on is given by:

MF_margined = (3/2) × √(MPOR / 250)

where MPOR is the Margin Period of Risk in business days and 250 is the number of business days in a year. The MPOR captures the time between the last received margin payment from a defaulting counterparty and the close-out and liquidation of the netting set, including the period required to replace the trade in the market.

Current Basel floor (CRE52.50)

Basel CRE52.50 applies a minimum MPOR of 10 business days for non-centrally cleared OTC derivative transactions. This floor reflects the supervisory assessment of the time required, under stressed market conditions, to identify a default, initiate close-out, dispute valuation, and liquidate positions across a range of bilateral settlement infrastructures and counterparty operational profiles. At MPOR = 10 business days, MF_margined = 0.300.

Existing carve-out: centrally cleared transactions (CRE54)

Basel CRE54.12 permits a clearing member to capitalise its exposure to clients on cleared transactions at an MPOR of at least 5 business days, reflecting the shorter close-out period on that bilateral leg; the clearing member's own trade exposure to the CCP remains subject to the 10-day floor of CRE54.8(2). The structural argument supporting this carve-out is that CCP intermediation reduces the close-out complexity and bilateral valuation disputes that drive the 10-day floor.

Forward-looking argument: onchain T+0 settlement

Onchain settlement with deterministic finality presents a stronger structural case for MPOR reduction than CCP clearing on three dimensions:

• Settlement finality. Atomic delivery-versus-payment onchain eliminates the settlement risk window that bilateral and CCP clearing both retain. Trades and collateral movements are either completed or not, with no in-flight state.
• Verifiable netting set state. The complete netting set composition, mark-to-market, and collateral state are observable onchain at any point in time. Reconstructing this state during close-out — a major source of the 10-day floor's conservatism — becomes instantaneous rather than days-long.
• Standardized valuation. When all counterparties to a netting set compute SIMM, SA-CCR, CVA, and exposure against the same calibrated curves using the same reference analytics, the valuation disputes that consume MPOR are eliminated by construction.

At MPOR = 1 business day, MF_margined = 0.0949, representing a 68.4% reduction in the SA-CCR PFE add-on relative to the current 10-day standard. The reduction flows through proportionally to SA-CCR EAD, BA-CVA capital charge (which scales linearly with EAD), KVA (derived from EAD), FVA (uncollateralized exposure window shrinks), and the analytical 97.5% PFE. ISDA SIMM Initial Margin is unchanged in this scenario: ISDA SIMM v2.8 is calibrated to a fixed 10-day liquidity horizon and does not vary with the settlement model.

Platform-agnostic framing

The MPOR reduction argument advanced in this section is not contingent on the adoption of any specific platform or vendor. The structural properties enabling sub-day MPOR — atomic settlement, deterministic finality, verifiable netting set state, and standardized valuation — can be provided by any qualifying onchain infrastructure. The argument is for supervisory recognition of the settlement model under CRE52.50, not for recognition of any particular provider.

Regulatory pathway

Sub-day MPOR recognition for non-cleared OTC derivative netting sets would require either (i) interpretive guidance from the relevant supervisor (FRB, OCC, CFTC for US banks; PRA, ESMA for European institutions; BCBS for international consistency) applying the existing CRE52.52 maturity factor formula at a sub-10-day MPOR for qualifying onchain netting sets; or (ii) a formal Basel consultation amending the 10-day floor in CRE52.50 to incorporate a graduated MPOR by settlement model. Neither pathway has been initiated as of the Calibration Date.

M.8CVA Methodology

Credit Valuation Adjustment is computed under the Basic Approach to CVA (BA-CVA), reduced version, per Basel MAR50.13. The reduced version applies to banks without eligible CVA hedges and is the formulation used by the vast majority of institutions without internal CVA risk-management approval.

Framework selection rationale

Basel MAR50 offers two approaches to CVA capital: (i) the Standardized Approach (SA-CVA, MAR50.27–50.77) requiring internal CVA risk-management approval and sensitivity-based computation; and (ii) the Basic Approach (BA-CVA, MAR50.13–50.26) applicable as the default for all other banks. Within BA-CVA, the reduced version (MAR50.14–50.16) excludes recognition of eligible hedges, while the full version (MAR50.17–50.26) incorporates single-name and index CDS hedge offsets.

This analysis adopts the BA-CVA reduced version because (a) most non-G-SIB banks do not have CVA hedging programs that qualify for full-version recognition, (b) the reduced formulation provides a like-for-like comparison across regimes without confounding hedge optimization, and (c) under the Basel 3.1 output floor (72.5%; BCBS phases to 1 January 2028, EU CRR3 Art. 465 to 2030), internal-model CVA charges converge toward standardized BA-CVA anyway.

Single-counterparty BA-CVA capital formula (MAR50.14 discount scalar, MAR50.15 SCVA)

K_BA-CVA = DS × (1/α) × S_c × M_eff,c × EAD_c × DF_c · DS = 0.65, α = 1.4

where:

• S_c = supervisory weight by counterparty rating bucket per MAR50.16 (see table below);
• M_eff,c = effective maturity per MAR50.15 (weighted cash-flow average);
• EAD_c = SA-CCR Exposure at Default per §M.7;
• DF_c = the MAR50.15 supervisory discount factor (1 − e^{−0.05·M_eff}) / (0.05·M_eff) — a fixed 5% supervisory constant and an integral average, not a point factor at the calibration risk-free rate.

Supervisory weights by counterparty rating (MAR50.16)

Effective maturity by reference product (MAR50.15)

The effective maturity M_eff under MAR50.15 is the weighted cash-flow average of trade payment dates. For vanilla products in this analysis:

Reference portfolio CVA (D2D, Regime A, IG counterparty)

Counterparty rating sensitivity

Switching counterparty rating bucket scales CVA proportionally to the supervisory weight ratio. For the same trade, the CVA capital charge against a HY counterparty is exactly 2.4× the IG charge (S_c 12% vs 5%), as it is against an Unrated counterparty — MAR50.16 Table 1 assigns both the same 12% weight. This is a structural feature of BA-CVA: the counterparty's credit quality drives the CVA capital regardless of trade economics, and the same derivative position with a worse-rated counterparty consumes proportionally more capital.

Calibration status

BA-CVA formula is exact per MAR50.13, MAR50.14, MAR50.15, and MAR50.16. Magnitudes are derived directly from §M.7 EAD calibrations. The supervisory weights are Basel-published; the calibration-date CDX index spreads (CDX IG 52.845 bp, CDX HY 312.320 bp) are referenced as market context for the bucket assignments, not as direct inputs (BA-CVA reduced does not use spread inputs, only the bucketed supervisory weight S_c).

M.8b · Crypto CVA — basic approach only

Corresponds to §M.15.2 of the methodology document.

SCO60.91 does not permit the standardised approach to CVA for derivatives referencing Group 2a cryptoassets. The basic approach is used throughout for the crypto set, on the same corrected formulation as above — including the 1/α divisor and the 0.65 discount scalar of MAR50.14. No crypto-specific parameter enters the CVA calculation beyond the SA-CCR exposure it is computed on (§M.7b) and the effective maturity, which for the perpetual is the margin period of risk rather than a contractual tenor.

M.9FVA Methodology

Funding Valuation Adjustment captures the cost of funding residual uncollateralised exposure at the bank's marginal funding spread over OIS. It is distinct from MVA (§M.9b), which is the funding cost on posted initial margin. The two are charges on different exposures and are neither netted nor substituted for one another.

Framework

Under the standard Burgard-Kjaer / Andersen-Pykhtin framework, FVA is obtained by integrating expected exposure net of collateral coverage over the trade life:

FVA = − ∫ s_f × EE(t) × (1 − c) × DF(t) dt

where:

• s_f = 75 bp (bank marginal spread over OIS, per §M.13 — indicative representative dealer);
• EE(t) = expected exposure profile, simulated on the XVA Engine tab;
• c = collateral-coverage factor by regime (0.70 nettable / 0.45 pre-eligibility / 0.00 non-nettable);
• DF(t) = risk-free discount factor at r = 4.1675% (§M.13).

The figure carried on the Capital Metrics tab is MVA, the funding cost on posted initial margin, because IM funding is the funding quantity that varies with collateral eligibility and therefore with the comparison this analysis makes. The XVA Engine tab presents FVA and MVA separately.

Why FVA is bounded under daily CSA

Netting-set note: the FVA and MVA figures in this analysis are computed on standalone per-trade SIMM IM. In a live netting set the funded IM is the netted amount, which is smaller than the sum of standalone amounts — the per-trade figures are therefore conservative (they overstate the funding cost) on this dimension.

Pre-CSA derivatives accounting historically computed FVA on the full uncollateralized expected exposure path — a large, time-extensive quantity. Under modern daily-margin CSAs with eligible IM and VM, the uncollateralized window is the close-out window (10 business days under CRE52.50), and the dominant FVA driver becomes the segregated IM funding gap rather than ongoing exposure funding. This produces materially smaller and more stable FVA than the legacy frameworks.

Reference portfolio MVA (Regime A baseline, daily CSA)

Regime variation under USDM1 — the precise mechanism

The classical FVA above is mechanically a function of (a) the bank's funding spread, (b) the residual uncollateralized exposure during the close-out window, and (c) trade duration. It does not include collateral yield as an input — collateral yield enters the XVA framework through MVA (§M.9b), not through FVA in the standard Burgard-Kjaer/Andersen-Pykhtin formulation. Industry shorthand sometimes describes the USDC/USDT “dead IM” effect as “FVA savings”; the underlying mechanic is more precisely an MVA effect compounded by a non-eligibility double-funding penalty:

• USDC/USDT scenario (Regime B): USDC/USDT is not enumerated as eligible IM under CFTC §23.156 (current state). A bank that holds USDC/USDT as an inventory or treasury position must additionally post separate cash or UST as IM. The bank therefore funds two positions simultaneously — USDC/USDT held at funding spread (USDC earns 0% to the holder per GENIUS Act §4(a)(11)), plus cash IM at funding spread (cash IM earns OIS at custodian). Aggregate funding cost = MVA on cash IM + double-funding penalty on USDC/USDT notional.
• USDM1 scenario (Regime D): If USDM1 entered the §23.156 enumerated eligible IM list, the same USDM1 a bank holds could be directly posted as IM. Accordingly, there would be no double-funding penalty, and a single instrument could serve as both the held position and the margin posting. The MVA on the IM posting would be computed under the asymmetric framework (§M.9b) and depends on USDM1 yield relative to bank funding cost.
• USDM1 vs cash IM (both eligible): there is no double-funding penalty in either case. The economic difference is the asymmetric MVA: cash IM MVA = funding_spread × IM × A (75 bp at calibration); USDM1 IM MVA = (funding_rate − USDM1_yield) × IM × A. At the Calibration Date USDM1 yield sits below the cash OIS rebate (~3.58%) — by ~185 bp at the current single rate (1.73%), ~75 bp under the intended registered rate (2.83%) — so USDM1 IM MVA exceeds cash IM MVA by ~185 bp × IM × A at the current single rate, narrowing to ~75 bp under the intended registered rate. USDM1 is worse on MVA than cash at calibration; the case for USDM1 vs cash rests on capital, leverage ratio, HQLA, and settlement-mobility dimensions, not on funding economics.

The collateral yield retained by the holder (the collateral carry dimension quantified in §M.12) is a separate retained-yield economic that flows to the holder's P&L as interest income, independent of FVA. The integrated economic position of a derivative + collateral package is therefore: trade P&L − XVA (CVA + FVA + KVA + MVA) + collateral carry. Carry offsets the total stack at the P&L level; it does not reduce FVA mechanically.

Calibration status

FVA formula and bank funding spread (75 bp over OIS, indicative representative dealer per §M.13) are stated transparently. IM amounts derive from §M.6 calibrations. The annuity factor uses risk-free rate r = 4.1675% (SOFR 5Y at the Calibration Date), consistent with §M.13 curves.

M.9bMVA Methodology — Margin Valuation Adjustment

Margin Valuation Adjustment captures the funding cost specifically attributable to Initial Margin (IM) posting under daily-CSA arrangements. While FVA in §M.9 above captures the general funding-spread differential on margined exposure, MVA isolates the IM-funding component for explicit attribution and for the forward-looking T+0 analysis.

Framework

Under modern UMR-compliant CSA arrangements, the bank posts IM in eligible collateral to the counterparty (or to a tri-party custodian). The bank funds this posted IM at its marginal funding spread above OIS, while the IM itself earns at most OIS. The economic loss accruing to the bank is the spread differential applied to IM held for the effective trade life:

MVA ≈ funding_spread × IM × A(M_eff)

The tool presents MVA as a linear approximation, consistent with industry practice for monitoring and pricing-desk views. A full Markovian path simulation of IM under stochastic risk drivers (Andersen-Pykhtin-Sokol 2017) is deferred — the linear approximation is appropriate for the bullet-pay structures in the reference portfolio, where IM trajectory closely tracks the deterministic profile of A(M_eff).

FVA/MVA decomposition

For derivatives with daily CSA and segregated IM, the conventional XVA framework decomposes the total funding cost as FVA (on the residual uncollateralized exposure window during MPOR) plus MVA (on the IM principal posted over trade life). The tool presents these separately in the XVA Engine waterfall to surface the IM-funding component explicitly — relevant for the forward-looking T+0 analysis below.

Asymmetric MVA framework — how collateral choice enters MVA

The cash-IM formulation above assumes posted IM earns OIS at the custodian (the standard tri-party rebate). For collateral types that earn at a different rate, the more general formulation nets the actual collateral yield against the bank's funding cost:

MVA = (funding_rate − collateral_yield) × IM × A(M_eff)

For cash IM at custodian (yield ~3.58% ≈ SOFR overnight at the Calibration Date, less small custodian spread), this reduces to the standard ~75 bp × IM × A.For UST coupon-bearing IM (~4.23%, H.15 1–3Y constant maturity 3-point average) - MVA is slightly lower than cash IM because UST yield exceeds the cash rebate.For BUIDL (3.49%, DefiLlama Institutional pool net of 50 bp mgmt fee): MVA is essentially equivalent to cash IM.For USDM1 yield (1.73% current single rate; 2.83% intended registered rate, H.15 1M Treasury constant maturity 3.730% − Spread) - MVA is higher than cash IM by approximately (3.58% − 1.73%) = 185 bp × IM × A at the current single rate (75 bp under the intended registered rate), because USDM1 yield sits below the cash rebate at the Calibration Date.

Implication for the USDM1 case - at calibration-date rates, posting USDM1 as IM produces ~185 bp higher MVA than posting cash at the current single rate (narrowing to ~75 bp under the intended registered rate).The argument for USDM1 vs cash IM is therefore not an MVA argument; it rests on the capital (SA-CCR EAD, CVA, KVA), leverage ratio, HQLA, and settlement-mobility dimensions documented in §M.7, §M.7.5, and §M.12.The USDM1 yield itself (the collateral carry) is retained by the holder as a separate economic and is logically additive to the XVA stack — it does not appear in MVA except via the asymmetric-framework offset above.On funding economics, so long as USDC/USDT is not §23.156-eligible IM, a USDC inventory position requires posting separate cash IM and creates a double-funding penalty. On the Capital Metrics tab that penalty is carried inside MVA, as the indicative regime multiplier disclosed at §M.7c (1.45× in Regime B, 1.10× in Regime C); the formula above, and the XVA Engine tab, carry no regime term and are invariant.

Forward-looking dimensions

Important distinction: Under ISDA SIMM v2.8, SIMM IM is calibrated to a fixed 10-day liquidity horizon and does not depend on the regulatory MPOR. Since MVA = funding_spread × SIMM_IM × annuity, MVA in the current ISDA SIMM framework does not scale with regulatory MPOR. The T+0 reduction visible on adjacent XVA cards (CVA, FVA, KVA, PFE) is a Basel SA-CCR effect (CRE52.52) and applies to those metrics only.

The forward-looking dimension that directly affects MVA is HQLA Level 1 recognition of USDM1. When USDM1 is used as posted IM, Level 1 classification means it counts at face value in the bank's LCR/NSFR composition rather than being excluded or haircut. This is a balance-sheet-level benefit on the funding side of the MVA equation rather than a reduction in the MVA charge itself: the bank funds the same IM principal, but the IM held also contributes to the LCR buffer at face value, materially improving the net economic position of the IM posting.Please see §M.12 HQLA Framework for a deeper analysis of the Level 1 / Level 2A architecture.

A second-order forward-looking scenario would be ISDA SIMM recalibration to align the SIMM horizon with T+0 settlement reality. If ISDA were to recalibrate SIMM to a 1-day horizon, SIMM IM (and therefore MVA) would compress by approximately 68% (the same √(1/10) scaling that applies to SA-CCR margin factor under CRE52.52). This is not current ISDA standard, is not reflected in the tool, and is noted here only for completeness.

Calibration status

Linear approximation per industry practice for bullet-pay derivatives. Funding spread (75 bp) and risk-free rate (r = 4.1675%) per §M.13. SIMM IM held constant under settlement-model transitions per ISDA SIMM v2.8 fixed 10-day horizon. Full Markovian IM-path simulation is outside the scope of this engagement.

M.9c · Crypto funding — MVA on prospective initial margin

Corresponds to §M.15.8 of the methodology document, which carries the fuller treatment.

The construction is unchanged — MVA = funding spread × IM × A(M_eff) — but the initial margin is the prospective crypto figure of §M.6b: $13,200,000 per $10M of notional, being 132% of notional, and not a SIMM output. Two consequences follow. Crypto MVA dominates the funding line of any book carrying dated crypto exposure, because that IM is an order of magnitude above the rates and FX products at equal notional. The perpetual is the exception, and not a small one: its effective-maturity convention collapses the annuity so far that its MVA of $3,957 sits below the IRS 5Y, the IRS 10Y and the XCCY 5Y. And the whole of it inherits the prospective status of the risk weight — it is no better established than the margin figure it funds.

Effective maturity separates the three products. The dated forward and the total return swap run to one year, giving A(1.0) = 0.9794. The perpetual takes the margin period of risk as its effective maturity, M_eff = 10/250 = 0.04, giving A(0.04) = 0.0400 — a factor of 24.5 lower. MVA falls with it, to 4.1% of the dated products’ on identical notional and initial margin.

This is a convention, not a market observation: it treats a position with no contractual maturity as spot risk held over the close-out window. The M_eff toggle exposes a one-year rolling convention and a floor treatment, and under the one-year convention the difference disappears entirely. Any comparison of perpetual against dated economics should establish which convention is in force first.

M.10KVA Methodology

Capital Valuation Adjustment is the expected present value of regulatory capital cost over the derivative trade's life. KVA reflects the bank's cost of equity capital consumed against the counterparty exposure, computed as the annuity present value of the capital amount times the bank's cost-of-equity hurdle rate.

Framework

The capital required against a derivative netting set at any time t under Basel III Standardized Approach is:

K(t) = EAD(t) × RW × 8%

where EAD(t) is the SA-CCR exposure path (§M.7), RW is the counterparty risk weight (CRE20.18/20.42/20.43), and 8% is the Basel III minimum total capital ratio (CET1 + Tier 1 + Tier 2). The annual cost of holding that capital is K × hurdle, where hurdle is the bank's cost of equity capital. Aggregating over the trade's effective life:

KVA = hurdle × ∫₀^M_eff K(t) × DF(t) dt ≈ hurdle × K × A(M_eff)

under the simplifying assumption that K(t) is approximately constant over the trade life (true for at-the-money interest rate swaps and short-dated FX forwards; for amortizing or aging products, K(t) declines, slightly reducing KVA below the analytical estimate).

Counterparty risk weights (CRE20.18/20.42/20.43)

Reference portfolio KVA (D2D, Regime A, IG counterparty; CCR component)

Hurdle rate

Cost-of-equity hurdle is fixed at 10% per §M.13 (industry-standard banking assumption, indicative for US mid-size and large banks). Banks with higher cost-of-equity assumptions (e.g., 12–15%) would observe proportionally higher KVA; banks with lower assumptions (8–9%) would observe proportionally lower KVA.

Counterparty sensitivity

KVA is not linear in RW, because it is the composite of two components that scale on different dimensions: the CCR leg scales by the risk weight (IG 50% → HY 100% → Unrated 150%), while the CVA leg scales by the supervisory weight S_c, which MAR50.16 Table 1 sets at 12% for both HY and Unrated against 5% for IG. Across the seven products and four regimes the composite ratio to IG runs 2.01–2.29× for HY and 2.57–2.99× for Unrated — never the 2× and 3× the risk weight alone would imply. This is a structural feature of the Standardized Approach to credit risk — the same derivative position with a worse-rated counterparty consumes proportionally more capital over the trade life.

Calibration status

KVA formula and hurdle assumption are stated transparently. EAD derives from §M.7. RW per CRE20.18/20.42/20.43. Calibration consistent with §M.13.

M.10b · Crypto capital — KVA on the SCO60.98 exposure

Corresponds to §M.15.8 of the methodology document, which carries the fuller treatment.

Both components are computed on the SCO60.98 exposure of §M.7b: KVA_CCR = EAD × RW × 8% × h × A(M_eff), and KVA_CVA = BA-CVA × h × A(M_eff), at a hurdle h = 10% and RW = 50% in the investment-grade context. No crypto-specific parameter enters beyond the exposure itself. The Group 2a classification enters through §M.7b; the 1250% Group 2b weight applies to held inventory (§M.12), not to counterparty exposure, and does not appear here.

The perpetual falls twice. KVA_CCR carries the annuity once and drops to 4.1% of the dated products’. KVA_CVA carries it twice — once through the BA-CVA it is computed on, which is itself linear in M_eff, and once through its own discounting — so it falls to 0.17%. The KVA this tool displays is the composite of the two, CCR + CVA: $220 against $8,247, or 2.7%. Compare components when comparing conventions — the composite moves with their mix, and neither component ratio describes it.

Potential future exposure does not follow this pattern: it is not annuity-weighted, and §M.11 computes it on a calendar clock from Bitcoin implied volatility. It is identical across all three products at $1,257,912 per $10M.

M.11PFE Methodology

Potential Future Exposure is computed on an analytical-approximation basis (not Monte Carlo). For margined OTC derivatives under daily CSA, the close-out exposure window is the dominant driver of counterparty risk, and the 97.5% PFE is well-approximated by a closed-form normal-quantile expression scaled by underlying volatility and risk sensitivity over the Margin Period of Risk.

Framework

PFE_97.5 = z_97.5 × σ_$_MPOR

where:

• z_97.5 = 1.96 is the one-sided 97.5% quantile of the standard normal distribution;
• σ_$_MPOR = σ_$_daily × √MPOR scales the daily exposure volatility over the close-out window;
• σ_$_daily = (risk sensitivity) × (underlying daily volatility) for each risk factor;
• MPOR = 10 business days under the Basel CRE52.50 floor for non-cleared OTC.

Underlying volatility inputs (Calibration Date)

Reference portfolio PFE (Regime A baseline)

Why analytical

This tool specifies PFE on an analytical-approximation basis explicitly to avoid Monte Carlo path simulation.For margined daily-CSA OTC derivatives, the close-out exposure distribution over a 10-day window is well-approximated by a normal distribution scaled by underlying volatility — the exposure path's complex middle is irrelevant because daily margin compresses the relevant exposure to the close-out window.The closed-form expression above produces a defensible 97.5% PFE without requiring Monte Carlo, while the XVA Engine tab provides the path-simulation view of the same metric; the reconciliation boundary between the two is set out at §M.14.

Calibration status

PFE formula and scaling are exact under the analytical framework. Volatility inputs are calibration-date indicative. The PFE magnitudes presented here are reconciliation targets for the XVA Engine tab.

Collateral regime. The formula above contains no collateral-regime term. On the Capital Metrics tab PFE nonetheless carries an indicative regime multiplier — 2.10× in Regime B, 1.12× in Regime C, 1.05× in Regime D pre-relief — representing the loss of mark-to-market offset during close-out when collateral is not recognised. It is an assumption rather than a derivation; see §M.7c.

M.12Balance Sheet, Liquidity & HQLA Framework

Two balance-sheet and liquidity dimensions supplement the core capital and margin analysis, addressing treasury and ALM audiences.

Basel leverage ratio impact

Leverage ratio treatment for derivative netting sets is computed under the Basel III leverage ratio framework (CRR Article 429c): the exposure measure equals the SA-CCR EAD — α × (RC + PFE), α = 1.4 — with no collateral credit against the PFE add-on and replacement-cost offset available for qualifying cash variation margin only. For the at-par, daily-margined reference book RC = 0, so leverage exposure scales with the CRE52.52 maturity factor and mirrors the §M.7 EAD treatment by regime: cash + UST and USDM1 (margined, Treasury-equivalent under the §M.4/§M.5 conditions) ≈ $9.3M per $1B notional on the 5Y IRS anchor; USDC/USDT under the §M.4 unenforceable-margin interpretation take unmargined treatment ≈ $28.8M; TMMF look-through ≈ $11.0M. The forward T+0 · 1-day MPOR state (§M.7.5) reduces USDM1's leverage exposure a further 68.4% to ≈ $2.9M, conditional on supervisory recognition. Click any row of the leverage table for the full derivation.

LCR HQLA classification — three-state framework

Liquidity Coverage Ratio classification under the Basel framework (BCBS 238, consolidated under Basel LCR20–LCR30) assigns collateral assets to: Level 1 (cash, central bank reserves, qualifying 0% risk-weight sovereign debt — 0% haircut, no cap on share of HQLA buffer); Level 2A (qualifying sovereign and supranational debt at 0%–20% risk weight — 15% haircut, included in the 40% Level 2 cap); Level 2B (qualifying corporate debt and equity — 25–50% haircut, within the 15% Level 2B cap inside the broader 40% Level 2 cap); or Non-HQLA (excluded from the LCR buffer entirely). Cash and US Treasuries receive Level 1 treatment unconditionally. USDC/USDT are private-issuer instruments not recognized under the LCR HQLA framework today.

For USDM1, the tool exposes two selectable states from the configuration bar — the Level 2A floor and the Level 1 ceiling — and shows the unrecognised case alongside them as a reference card on the Balance Sheet tab. All three are set out below:

Not recognized — the conservative default state absent supervisor guidance. USDM1 contributes nothing to the LCR buffer. This is where an institutional compliance team starts.

• Level 1 HQLA — USDM1 treated as equivalent to direct US Treasury holding via sovereign look-through to the underlying US Treasury collateral (held by the trustee under perfected first-priority security interest). 0% haircut, no Level 2 cap.
• Level 2A floor — USDM1 treated as a government/agency-equivalent secured sovereign instrument: 15% haircut, included within the 40% Level 2 cap.

Level 1 case is grounded in three primary pillars:(i) the legal structure (perfected first-priority security interest in pledged US Treasuries via the US trustee);(ii) the Federal Reserve Discount Window collateral framework explicitly accepts investment-grade-rated Brady Bonds as eligible collateral (Collateral Eligibility framework, last updated December 1, 2025), with USD-denominated short-to-medium-duration Brady Bonds receiving margins of 96–98% — approximately 1% below comparable US Treasuries per the Collateral Valuation table (eff. July 1, 2025); and(iii) the convergence of public industry signals (the RMI Banking Commission directive of November 11, 2025; the Fed/OCC/FDIC interagency FAQ of March 5, 2026 confirming technology-neutral capital treatment of eligible tokenized securities; and the structural analysis of issuer and independent regulatory counsel).

Three important framework distinctions:

1. "Sovereign look-through" applies the LCR framework’s own conditions for indirect and structured holdings (BCBS 238; June 2017 LCR FAQs) — transparency to the underlying pool, a perfected claim on the underlying free of legal or operational impediments, and no redemption gates or suspensions — to a tokenized sovereign-collateralized instrument.

Under BCBS 238 fn. 15, a 0% risk weight is assigned at national discretion under Basel II para. 54. The RMI directive’s stated basis maps to the domestic-sovereign Level 1 categories, which binds domestically.

For non-RMI institutions, Level 1 argument is the look-through to the 0%-weighted US Treasury collateral.

2. Federal Reserve Discount Window eligibility is a distinct regulatory framework from Basel HQLA Level 1 classification. (Discount Window is the Fed's lender-of-last-resort facility, Basel HQLA is bank liquidity-buffer composition under LCR/NSFR.)

USDM1’s Brady Bond precedent supports the structural analogy for HQLA recognition, but does not directly determine it.

3. As noted above, the RMI Banking Commissioner — the domestic supervisor — has issued a formal directive (November 11, 2025) instructing supervised institutions to treat USDM1 as Level 1 HQLA with a 0% sovereign risk weight.

Outside the RMI, no prudential supervisor has issued specific guidance and classification is determined institution-by-institution by the relevant prudential regulator (Fed/OCC/FDIC for US banks, ECB/PRA for EU/UK banks).

Operational pathway considerations

Structural eligibility is one dimension.

Operational pledging mechanics are another. The Federal Reserve Discount Window framework requires intermediated securities to be pledged through Depository Trust Company (DTC), Euroclear, Clearstream, or the Federal Reserve's Fedwire Securities Service (per Acceptance Criteria for Securities, Fed Discount Window Collateral Eligibility framework).

USDM1, as a natively onchain instrument, settles outside of these traditional intermediated-securities depositories at present. The structural argument for regulatory recognition (perfected first-priority security interest, sovereign issuer, US Treasury collateral) is distinct from the operational pathway required for actual collateral posting under specific regulatory facilities. Bridging the operational gap typically follows recognition rather than preceding it.

USDM1 is a member of institutional working groups actively engaged on this operational dimension — including the Canton Network industry working group (Bank of America, Citadel Securities, Cumberland DRW, Société Générale, Tradeweb, Virtu Financial) and multi-month DTCC pilots; see the program references in the source list.

On distribution and custody, USDM1 is offered through Anchorage Digital, an OCC-chartered digital bank, and through BitGo Bank & Trust, National Association, an OCC-regulated digital asset trust bank. It can also be held through tZERO Digital Asset Securities, LLC, an SEC-registered, FINRA-member broker-dealer custodian of digital asset securities (announcement, June 4, 2026) — three independent regulated custody rails — with deposit and withdrawal support at Bank of Guam, an FDIC-insured US bank.

USDM1 is issued and supported across multiple blockchain networks (Stellar, Ethereum, Solana).

For LCR/NSFR purposes, the operational gap is less binding because LCR HQLA composition is an internal balance-sheet measure rather than an active-pledging facility — recognition translates more directly into balance-sheet treatment than into pledging mechanics.

Supervisory acceptance pathway

HQLA classification is ultimately a determination made by each prudential regulator with jurisdiction over the holding institution — not by issuers, counsel, or industry consortia.

For US banks, the primary federal regulators are the Federal Reserve (state member banks and bank holding companies), the Office of the Comptroller of the Currency (national banks and federal savings associations), and the Federal Deposit Insurance Corporation (state non-member banks and state savings associations).

For European banks, the regulators are the European Central Bank (significant institutions under the Single Supervisory Mechanism) and national supervisors (less-significant institutions and EU member states outside the eurozone).

For UK banks, the regulator is the Prudential Regulation Authority.

In practice, institutional compliance teams will typically: (i) document USDM1’s legal structure (security interest, perfection, custody chain); (ii) obtain external counsel opinions on classification defensibility; (iii) engage their prudential supervisor in pre-clearance discussions for material classification questions; (iv) document the analytical basis internally; and (v) apply the classification subject to ongoing supervisory review.

Closing analytical summary

Level 1 HQLA classification is supported by a coherent legal-structural argument (perfected first-priority security interest in US Treasury collateral) and a coherent regulatory-precedent argument (Fed Discount Window treatment of analogous Brady Bond structures), but no supervisor outside the RMI has yet confirmed it for USDM1 specifically.

Absent any classification decision, the tool’s user can toggle between states to understand the capital and liquidity consequences of each classification under each user's own institutional risk-policy framework. This tool is not intended to advocate for a particular classification.

Collateral carry framework

Collateral carry is the annualized yield retained by the holder of a collateral asset while the asset sits on balance sheet or is posted under a bilateral CSA.The carry dimension is logically independent of the capital, leverage ratio, and HQLA dimensions analyzed above — an asset can be capital-efficient and yield-positive (cash, UST), capital-efficient and yield-zero (USDC/USDT), or capital-efficient and yield-bearing under a different mechanism (USDM1). The Balance Sheet Carry panel quantifies the calibration-date yield across a comparator set.

USDM1 yield mechanism (per Indenture dated September 23, 2025). USDM1 is structured as a perpetual adjustable-rate secured bond. Interest accrues daily at the Treasury Rate minus a Spread, capitalized monthly.

The Treasury Rate is anchored to H.15, the Federal Reserve's Selected Interest Rates statistical release, using the one-month US Treasury constant maturity rate, reset on the third Wednesday of each month.

Spread changes require thirty (30) days' written notice to the Trustee, Transfer Agent, and Token Agent.

Currently a single 2.00% Spread applies to all holders.

In a potential future state, a 90 bp Spread may be introduced for registered bondholders with permissionless holders paying an additional 100 bp Compliance Surcharge (total 190 bp), reflecting the higher monitoring cost of unregistered wallets under the Republic's AML obligations. Potential Future Yield in Asset Configuration presents the potential impacts of this structure.

At the Calibration Date, the H.15 1M Treasury constant maturity rate (the rate the Indenture explicitly anchors to) is 3.730%.Under the single-rate structure, all-holder yield = 3.730% − 2.00% = 1.73%.Under a potential future structure, registered yield = 3.730% − 0.90% = 2.83%; permissionless yield = 3.730% − 1.90% = 1.83%.Note that this differs from the 1M Treasury bill secondary-market rate (3.59%) by ~10 bp because the constant maturity rate is interpolated from the broader Treasury yield curve while the bill rate reflects the actual market yield on a specific bill close to maturity.USDM1's Indenture explicitly references “the one-month US Treasury constant maturity rate” with the constant maturity figure as the binding input.

Why USDC/USDT pay 0% to holders. Section 4(a)(11) of the GENIUS Act (S.1582, 119th Congress) provides: “No permitted payment stablecoin issuer or foreign payment stablecoin issuer shall pay the holder of any payment stablecoin any form of interest or yield (whether in cash, tokens, or other consideration) solely in connection with the holding, use, or retention of such payment stablecoin.” Issuers retain the income on backing assets (predominantly short-duration US Treasury bills). The stablecoin holder bears the opportunity cost of cash sitting idle while the issuer captures the T-bill carry.USDM1 is a sovereign bond structurally outside this prohibition, as the issuer (GRMI) is itself the sovereign and the instrument is a debt obligation rather than a payment stablecoin. The interest mechanism is part of the bond's economic terms by Indenture design.

Carry comparison - USDM1 provides a carry advantage against USDC/USDT which does not provide yield. Against cash IM at a custodian (~3.58% at the Calibration Date, OIS-based tri-party rebate given SOFR overnight 3.66%), USDM1 yield sits below by approximately 185 basis points at the current single rate (1.73%), or 75 basis points under a potential future registered rate (2.83%).Against US Treasuries held outright (~4.23%, H.15 constant maturity 3-point average of 1Y 4.08% / 2Y 4.28% / 3Y 4.34%) or BUIDL (3.49%, DefiLlama Institutional pool, net of BlackRock 50bp management fee), USDM1 is calibration-date 176–250 bp negative on yield at the current single rate, and 66–140 bp under a potential future registered rate.The case for USDM1 against cash, UST, and BUIDL therefore does not rest on carry alone — it rests on the combination of carry retention plus the netting, capital, leverage ratio, HQLA, and settlement-mobility dimensions documented elsewhere in this section.

BUIDL (BlackRock USD Institutional Digital Liquidity Fund) is the closest tokenized alternative to USDM1 and merits a direct comparison.

The two instruments look similar on the surface (onchain, UST-backed, institutional collateral target) but are legally and structurally different.

BUIDL is a fund interest under Reg D 506(c) with a 50 bp management fee and $5M minimum subscription, custodied by BNY Mellon and tokenized by Securitize.

USDM1 is a secured sovereign bond under a Marshall Islands Indenture governed by NY law.

The regulatory consequences of this distinction are material.

Under SA-CCR CRM, BUIDL receives partial recognition via the fund look-through provisions of CRE22.49/22.50 (~1.18× EAD multiplier); with a path to eligibility under UMR, USDM1 receives full sovereign CRM under CRE22.34(3)(a)(i) (1.00× EAD).

Under Basel LCR, BUIDL classification ceiling is Level 2A (15% haircut, included in the 40% Level 2 cap); USDM1 receives a Level 1 ceiling via sovereign look-through to the underlying US Treasury collateral.

Under CFTC §23.156, the MMF enumeration that would cover BUIDL is narrow; USDM1’s potential UMR eligibility would represent bringing tokenized sovereign instruments onto the eligible IM list.

USDM1’s bankruptcy safe-harbor treatment under US Bankruptcy Code §§555–561 is cleaner for securities-entitlement-based sovereign instruments than for fund-interest pledging arrangements.

The carry give-up captures these regulatory differentials in price.At the Calibration Date, USDM1 yield sits below BUIDL distribution — approximately 176 basis points at the current single rate (1.73%), 66 basis points under the potential future registered rate (2.83%) — relative to BUIDL distribution (3.49%), translating to ~$17.6M of yield per $1B held per year at the current rate, or ~$6.6M under the potential future registered rate.For a dealer where capital and liquidity constraints bind — the typical institutional buyer of derivative-margin collateral — the EAD reduction (from 1.18× to 1.00×), the HQLA Level 1 buffer expansion, the direct §23.156 IM eligibility, and the cleaner safe-harbor pathway collectively dominate the carry give-up by margin that varies with the bank's balance sheet composition.For a treasury where idle-cash carry is the dominant metric, BUIDL retains a small yield advantage at the Calibration Date.

Calibration source values (7/31/2026). Cash custodian rebate (~3.58%) is indicative of standard tri-party arrangements at SOFR overnight (3.66% per industry-standard composite USD SOFR curve) less ~5–10 bp custodian spread; institution-specific arrangements will differ.UST 1–3Y yield (~4.23%) is the 3-point average of Fed H.15 Treasury constant maturity 1Y 4.08% / 2Y 4.28% / 3Y 4.34% on 31 July 2026.BUIDL distribution rate (3.49%) is the Institutional share class observed on DefiLlama at the Calibration Date, with $887M TVL and 30d average APY 3.57%, net of BlackRock's published 50 bp management fee on the Ethereum BUIDL share class.USDM1 yields are derived deterministically from the Indenture mechanism applied to H.15 1M Treasury constant maturity rate 3.730%.

Poster's perspective

The symmetric analytical view from the collateral poster (buy-side) reflects margin credit per dollar of asset posted, eligibility status under standard institutional CSA terms, and the effective carry economics achievable by combining margin credit with yield retained on the posted asset.CSA haircuts under standard institutional terms reduce the margin credit per dollar posted (0% for cash, 0.5–2% for US Treasuries, 0–2% for tokenized assets eligible under the CSA, N/A for non-eligible private stablecoins).The poster panel frames USDM1 as Treasury-equivalent on both sides of the trade - the receiver gets the same regulatory capital and margin treatment as cash + UST; the poster receives the same CSA credit per dollar posted, plus retained Treasury-anchored yield.The Net effective $ / $1 / yr column combines margin credit and one-year retained yield to surface the integrated economic outcome of each collateral choice.USDM1 produces a positive net economic outcome ($1.012 at the current single rate, $1.023 under a potential future registered rate) against UST ($1.037) and BUIDL ($1.015); USDC/USDT alone among the comparator set produces $0.000 because neither dimension contributes positive value.

M.13Market Data & Calibration

This tool is calibrated against market data captured at the Calibration Date (7/31/2026). Market data inputs are sourced from industry-standard market data composite, mid values, ACT/360 day count conventions for USD interest rate instruments.

Calibration snapshot

SOFR OIS curve — full tenor grid

Par swap rates (mid) across standard tenors used for IRS pricing and analytical-curve construction:

User overrides

The XVA Engine tab exposes the trade economics, counterparty credit, funding and capital inputs above as live controls, so a reader can re-run the simulation against their own assumptions. The Capital Metrics tab reads the calibration values directly and is not user-adjustable.

Source attribution

USD SOFR OIS, EUR OIS ESTR, EUR/USD spot and forward points, USD/EUR cross-currency basis, USD swaption ATM normal volatility surface, and CDX Investment Grade 5Y CDS reference are sourced from industry-standard market data composite, mid values, as of close 7/31/2026. The institutional composite curves available through CME, ICE, LCH, BGN, and ICAP/TP all converge to within approximately 1–2 basis points on the SOFR belly under standard bootstrapping conventions; the values used here are representative of any of these sources.

Collateral yield calibration sources (per §M.12 Collateral carry framework):

• USDM1 yield mechanism: USDM1 Indenture dated September 23, 2025 (drafted by Cleary Gottlieb Steen & Hamilton LLP, issuer’s counsel — structuring under New York law with partners in sovereign debt, UCC and secured transactions, creditors’ rights, netting and digital asset market infrastructure; public program announcement, June 4, 2026)
• Program references: Office of the Banking Commission, Republic of the Marshall Islands — directive on USDM1 to supervised institutions (November 11, 2025); Surus Trust Company announcement — US trustee, custodian and collateral agent for USDM1 (surus.io); USDM1 availability on Anchorage Digital (PR Newswire); Canton Network industry working group on onchain US Treasury financing (PR Newswire); “Treasuries On-Chain: An Industry Case for Change”, VX, January 2026; Federal Reserve / OCC / FDIC, “Capital Treatment of Tokenized Securities — Frequently Asked Questions”, March 5, 2026 (federalreserve.gov); tZERO Group: “tZERO Adds Sovereign Digital Bond Collateral to Regulated Broker-Dealer Custody”, June 4, 2026 (wire; also tzero.com newsroom).
• H.15 Treasury constant maturity rates: Federal Reserve Board H.15 Statistical Release (1M 3.730% at the 15 July 2026 third-Wednesday Indenture reset; 1Y 4.08%, 2Y 4.28%, 3Y 4.34% at 31 July 2026). Used for USDM1 yield derivation (1M) and UST 1–3Y reference (midpoint).
• SOFR overnight: industry-standard composite USD SOFR curve, 31 July 2026, 1WK point 3.66%. Used as proxy for cash custodian rebate basis.
• BUIDL distribution: DefiLlama BlackRock BUIDL pool tracker, Institutional share class (~$887M TVL), APY 3.49% / 30d average 3.57% at Calibration Date. BlackRock BUIDL management fee 50 bp (Ethereum/Arbitrum/Optimism share classes) per BlackRock press release November 13, 2024.
• GENIUS Act §4(a)(11) interest prohibition: Guiding and Establishing National Innovation for US Stablecoins Act, S.1582 (119th Congress, 2025–2026), as enacted.

GlossaryDefined Terms

Comprehensive reference of all defined terms used in the analysis. Hover over any underlined term in the tool to see a short definition with a link back to this glossary.

M.14Derived Presentation Metrics & Scenario Constructions

Numbering follows the methodology document. This tab carries §M.14.1, §M.14.4 and §M.14.6; the document additionally carries §M.14.2 (headline attribution decomposition), §M.14.3 (multi-scenario XVA parameterization) and §M.14.5 (perpetual funding convention).

M.14.1 · Per-trade spread internalization — cost stack and the IM operational fee

The all-in cost compared across regimes comprises the SA-CCR EAD capital snapshot (EAD × RW × 8%), MVA, composite KVA (CCR + CVA capital), and an initial-margin operational fee:

This covers segregation, tri-party custody and reconciliation. The IM principal is deliberately excluded: posted collateral is recoverable at trade end and is not a cost. Because SIMM IM is regime-invariant (§M.6), the ops fee is identical on both sides of every comparison and contributes exactly zero to the differential; it is carried so the denominator represents a complete all-in cost rather than a partial one.

M.14.4 · XVA Engine exposure construction

The XVA Engine's exposure profile is a Monte Carlo simulation of the product's own payoff. EE, ENE and the upper quantile are read from the simulated distribution of the trade's mark at each step, so each product carries the exposure shape it actually has rather than a shape shared across the book.

Revision note. An earlier build of this tab used a single closed form, EE(t) = N × σ × 5.0 × s_mult × √(t(T−t)/T) × m_regime, applied to every product, with ENE(t) = −0.92 × EE(t) and an upper quantile of 2.35 × EE(t). That construction had three defects. It amortised every product to zero at maturity, which is right for a par swap and wrong for a bullet — an FX or Bitcoin forward settles in one exchange at T, so its exposure peaks there. It returned identical exposure for the IRS 5Y and the XCCY 5Y, because the shape saw only tenor and both read the same rate volatility, ignoring the maturity notional exchange that distinguishes a cross-currency swap. And it drove the FX forward from a basis-point rate volatility scaled by a duration proxy of 5.0, a quantity unrelated to how an FX forward moves. Figures on this tab changed for every product except the IRS 5Y; §M.14.6 records the effect.

Simulated risk factors

Two factors are diffused on a common time grid, seeded so that every run is reproducible:

The short rate follows Hull-White single-factor with θ set to the Calibration Date curve level; the price factor is a driftless lognormal under the pricing measure. σ_r is the swaption-implied rate volatility (89.48 bp) and σ_S is the relevant price volatility: EUR/USD 1-month ATM implied at 5.165% for the FX and cross-currency products (§M.11 uses the same figure), and the BTC ATM implied term structure for the crypto products (§M.2).

Payoffs

A(·) is the annuity at the Calibration Date discount rate and df(·) the corresponding discount factor. Every payoff is linear in notional, so the simulation runs per unit notional and notional is applied afterwards as a scalar.

Profile statistics

At each step the simulated marks give the three series directly:

ENE is no longer a fixed multiple of EE: the ratio now falls out of the simulation and ranges roughly −0.84 to −1.00 across the reference portfolio. The upper series is a genuine empirical quantile rather than a scaled EE; as before it is not expected positive exposure, and the 1.19 factor rescaling it to the 97.5th percentile, Φ⁻¹(0.975)/Φ⁻¹(0.95) = 1.96/1.645, is unchanged. 2,000 paths are run over 60 steps; the previous 500 left roughly 25 observations in the 95% tail, too few for a stable quantile.

The collateral-mitigation factor m_regime (0.30 nettable / 0.55 pre-eligibility / 1.10 non-nettable) and the settlement multiplier s_mult (1.0 standard / 0.316 at T+0) apply to the simulated series exactly as before.

Level calibration

Shape and the EE/ENE/quantile relationship are simulated; the level is calibrated. A single scalar is applied to every product, fixed so that peak EE on a margined $10M IRS 5Y reproduces the institutional fair-value-equivalent magnitude carried since Phase 1 — ~$150,000 at the Calibration Date volatility and standard settlement. It is computed once from that reference case at the calibration-date inputs and then held, so moving the volatility slider still moves exposure rather than being absorbed by the constant. Because the same scalar applies throughout, the relative levels across products are the simulation's output, not an assumption: the crypto/rates exposure gap reflects a 42% price volatility against an 89 bp rate volatility.

with M_e(t) = max(0.5, (T−t)/2) the declining effective maturity, and DS = 0.65 per §M.8. The α appears in the CCR leg (converting EEPE to EAD) but not in the CVA leg: SCVA = (1/α) × RW × M × EAD × DF with EAD = α × EEPE, so the two cancel (MAR50.15 fn. 4). Applying 1/α again would double-count.

M.14.6 · Effect of the exposure revision

Peak EE per $1B notional at the default scope (D2BS, unrated counterparty, netting eligible, T+0 settlement), before and after the revision:

The IRS 5Y is unchanged because it is the calibration anchor. CVA, DVA, FVA, KVA, PFE and the scenario waterfall on the XVA Engine tab move with the profiles above. Capital Metrics, Margin Tools, Balance Sheet and Comparison are unaffected — those tabs read the supervisory-formula values in §M.2, which do not depend on the simulation.

M.18Limitations & Forward Extensions

The reference portfolio comprises seven illustrative non-cleared OTC derivative products — four rates and FX, three Bitcoin. Crypto initial margin is prospective: published ISDA SIMM has no crypto risk class, and the weights used here come from Federal Reserve staff research rather than from an ISDA calibration (§M.6b). ETH, XRP and SOL are carried in the calibration structure but not populated, and are outside the current scope.Additional product types — including equity derivatives, credit derivatives, commodity derivatives, inflation derivatives, structured products, exotic options, and centrally cleared trades — are outside the scope of the current tool and may be addressed in future iterations.

The analysis reflects regulatory frameworks as currently understood. Subsequent changes in regulatory interpretation, supervisory guidance, or formal rule revisions are not reflected absent a recalibration date.No representation is made that USDM1 will achieve UMR eligibility or relief on any particular timeline or in any particular form.

AddendumPhase 2 Methodology — Where It Lives

Sections §M.1–§M.14, §M.18 and the Glossary appear in this tab and in the standalone methodology document under the same numbers. The Phase 2 sections are delivered in that document only: the Phase 2 methodology is issued as a separate addendum, and carrying a second copy here would create a source that can drift from it. Each is implemented in the tool by a tab rather than by a section.

The crypto material in this tab sits as lettered subsections inside the sections whose frameworks it extends — §M.6b under SIMM IM, §M.7b under SA-CCR EAD, §M.8b under CVA, §M.9c under MVA and §M.10b under KVA — each carrying its correspondence to §M.15. That placement is deliberate: the crypto risk class and SCO60.98 are extensions of those two frameworks, not a separate methodology.