Corporate model: trading firms
This framework rates operating entities whose credit risk is dominated by probability of default, and then rates each facility extended to them separately. It is written for trading firms and market makers, and generalises to lenders, operating corporates and asset managers through class-specific calibration of the financial axis.
Framework Architecture
Five layers. Classify, anchor, modify, cap, then rate the facility.
| Layer | What it does |
|---|---|
| Layer 0 · Classify | Identifies the legal entity that is the obligor, tests whether it is bankruptcy-remote (if it is, the assessment routes to a structured methodology), assigns an obligor class, and locates the entity in its group structure |
| Layer 1 · Anchor | Crosses a business-risk axis with a financial-risk axis to produce a starting rank in probability-of-default terms |
| Layer 2 · Modifiers | Five independent reads off the same anchor: liquidity, financial policy, information quality, group support, and a bounded committee overlay, applied as notch adjustments and summed rather than chained |
| Layer 3 · Caps | Each cap is a worst-achievable outcome. The binding cap overrides the mechanics |
| Layer 4 · Facility | Estimates expected recovery on the specific claim and converts it to notches off the issuer rating. Run once per facility, not once per obligor |
| Layer 5 · Monitoring | One-directional: it can cut a rating, never raise one |
Anchor
Business risk crossed with financial risk gives the anchor rank. Seven inputs: four judgment scores on the business axis, three ratios on the financial axis. Each axis is scored from strongest to weakest and the two are read off a matrix that is monotone in both axes.
Business risk
| Sub-factor | What it captures |
|---|---|
| Scale and competitive position | Absolute capital and revenue scale, market position, franchise durability |
| Revenue quality and diversification | Recurring versus episodic revenue, strategy and counterparty concentration, single-venue dependence |
| Operating environment | Sector cyclicality, jurisdiction, quality of the supervisory and insolvency regime |
| Management, governance and risk control | Independence of risk from revenue, key-person concentration, ownership structure, incident history |
Management and governance carries the heaviest weight of the four, deliberately. For unsecured lending to trading firms, governance and control failure is the dominant tail rather than leverage.
This axis is one of only two places in the framework where analyst judgment is permitted.
Financial risk
Three dimensions, one primary metric each. The skeleton is class-agnostic; the metric and its thresholds are set by the obligor class.
| Dimension | Question it answers | Metric |
|---|---|---|
| Leverage and capital | How much loss does the balance sheet absorb? | Adjusted equity to adjusted assets |
| Earnings and coverage | Can the entity service its debt and rebuild the buffer? | Net income to equity, multi-year average |
| Buffer volatility | How fast can the buffer be consumed? | Maximum twelve-month drawdown to equity |
Buffer volatility is the load-bearing generalisation. It carries the substance of what a structural default model reaches for.
Adjusted equity and assets means net of intangibles, goodwill, related-party receivables, self-issued token holdings and restricted capital, after liquidity haircuts. That adjustment is where most of the real analysis sits, and it is shown as working rather than buried in a single figure.
Secondary metrics are documented rather than modelled, gross notional to equity, worst-month loss, mark-to-model share and counterparty concentration for trading firms; provision coverage, borrower concentration and weighted LTV for lenders. A secondary metric that materially contradicts the primary one is grounds for a logged overlay, never a silent adjustment to the input.
Modifiers
Liquidity and funding
This modifier evaluates the borrower's capacity to make the payment inside the cure window under stress, solvent or not.
It leads on a stress-coverage ratio: stressed liquidity available in the cure window against stressed obligations falling due in it, with both legs computed under a single consistent shock. The single-shock requirement is the entire design point.
The shock that threatens the payment is the same shock that fires margin calls and makes cash hard to raise. The numerator shrinks and the denominator inflates simultaneously. This is where delta-neutral books get caught: net-flat but grossly large, so a market move generates large calls on the losing leg while the offsetting gain is only available if the positions are enforceably cross-margined at the same venue.
Financial policy and trajectory
The leverage metric says where leverage is; this modifier says where it is going and who decides. Two firms can present identical ratios and be different credits: one having deleveraged into its current position, the other at the same level on the way up while funding a distribution to its owner.
It reads direction rather than level: leverage and equity trajectory over several periods; distribution and extraction policy, which for owner-managed firms is the dominant term because retained earnings are loss-absorbing capital and extracted earnings are not; stated intent against revealed behaviour through a bad period; growth appetite and how it is funded; and completed capital events rather than planned ones.
Information quality and verification
The framework's principal differentiator. Bidirectional, additive and multi-path: there is no single mandatory signal, and no single missing document is fatal. The premise is that the borrower is not the object of trust: their auditor, regulator, administrator and the chain are.
Group and external support
This is where structural subordination enters at issuer level. A holding-company claim is residual on operating-company equity and therefore ranks behind every operating-company creditor, trade payables included. A senior unsecured holdco note can be economically more junior than an opco subordinated loan. The question to ask of every claim is which legal entity issued it and what assets that entity actually owns.
Committee overlay
The named, bounded and auditable home for discretion: written rationale, peer or committee validation, logged. Deliberately narrow, and narrower than in earlier versions of the framework because the information-quality modifier is bidirectional and no longer the only route to an upgrade.
The rate and direction of overrides is itself monitored. The pressure on an override is structurally always upward; if most overrides are upward, the model has become a rubber stamp.
Caps
Caps bound what the mechanics can produce. The final rank is the worse of anchor-plus-modifiers and every applicable cap. Caps include but are not limited to: Information floor, scale floor, jurisdiction, integrity, sector.
The facility layer
One issuer, many facilities. The issuer credit rating is the starting point; recovery, seniority and structure notch it per instrument.
Estimating recovery
For an unsecured claim, recovery is the residual after claims senior to it, shared pro rata across claims of equal rank. For a secured claim, collateral is applied first and the shortfall then ranks unsecured.
Two inputs carry the analysis. Stressed value is adjusted assets after liquidity haircuts for a balance-sheet obligor, or stressed operating earnings at a distressed multiple for an operating corporate. Collateral haircuts must reflect stressed liquidation rather than a market quote: price shock, slippage at size, and the time and cost of enforcement in the relevant jurisdiction.
The subordination taxonomy
Four distinct things, routinely conflated. All are facility-layer inputs except structural subordination, which also enters at issuer level through the group modifier.
| Type | Mechanism |
|---|---|
| Contractual | Intercreditor agreement: the junior lender agrees to be paid after the senior. Same entity, same assets |
| Structural | Debt at holding-company level rather than at the operating company. Holdco claims are residual on opco equity |
| Temporal | Maturity profile: shorter-dated debt effectively exits first |
| Effective | Secured versus unsecured on the same contractual tier |
What is actually read: priority of payment in the relevant insolvency regime; guarantee coverage and whether guarantees run upstream, downstream or cross-stream; the collateral package and its exclusions; negative pledge and permitted-liens capacity; restricted-payment capacity; and any structural leakage route, including unrestricted subsidiaries and asset-transfer capacity of the kind seen in recent liability-management exercises.
Monitoring
Quarterly review at minimum, and more frequently below investment grade. A live breach such as assets dropping, coverage collapsing, a connection going dark, divergence from the reported book, or covenant and NAV-trigger proximity opens a review on either of two grounds:
- the risk worsened, or
- trust in the reported information decreased.
Either can cut a rating; neither can raise one. The separation matters: "the world got worse" and "your numbers cannot be relied on" are different findings, and both are legitimate downgrade causes.