Skip to main content
Alterscope rates tokenized real-world assets (tokenized Treasuries, money-market funds, stablecoins, private credit) with the same risk-factoring engine that scores DeFi instruments, extended by two RWA-specific factors. One unified model, two markets, one published methodology. The canonical output is a continuous 0–100 score; the AAA–D letter grade is a derived, monotonically-mapped view for readers who think in notches. What sets the methodology apart from incumbent RWA raters:
  • Continuous, not quarterly. Ratings recompute on a five-minute cadence from live on-chain and attestation data, not from a quarterly PDF cycle.
  • Factor-decomposable. Every score replays into its 9 factor contributions, and every factor into named subfactors with citations to the underlying data row (NAV snapshot, custody-attestation hash, audit-firm signoff event). No black box.
  • Bi-temporal and verifiable. Every rating row carries both validity time and transaction time, so you can ask “what was the rating as known on May 1?” and get a historically-honest answer. Pass asof on any ratings endpoint to replay the past.
  • Cascade-graph-aware. The Systemic factor measures inverse contagion — how many downstream positions degrade if this issuer drops notches — by querying Alterscope’s dependency graph, which is itself bi-temporal and auditable.

​
The 9-factor model

DeFi instruments use the existing 7 factors at their existing weights, with zero weight on the two RWA-specific factors. RWA instruments use all 9 with RWA-tuned weights. The weight vector is fixed per instrument type and published per methodology release. Factors 8 and 9 are the RWA-specific additions. The on-chain-derived Custody factor (17%) is deliberately weighted above the catalog-derived Issuer Governance factor (13%): attestations refresh continuously, while the issuer catalog is curated and refreshed quarterly, so the fresher signal carries more weight.

​
Factor 8 — Custody Attestation Quality (17%)

Four subfactors totalling 100 points; explicit, auditable, monotonic:

​
The Systemic factor — inverse contagion

The Systemic factor does not measure this instrument’s exposure to others. It measures how much of the ecosystem depends on it: the rating engine queries the cascade graph for downstream positions that would degrade if the issuer dropped notches, and scores the exposure-weighted sum against the instrument’s TVL. Because the graph is bi-temporal, the Systemic score is auditable — you can replay exactly which downstream markets drove a score on any past date.

​
Score → grade mapping

The 0–100 score is canonical; the letter grade is a published, immutable, monotonic mapping per methodology release. BBB− (50) is the investment-grade floor. Every rated instrument carries both surfaces at once — for example (score: 84, grade: "AA-", investment_grade: true).

​
Insufficient data is disclosed, never guessed

When a factor lacks signal it is marked not_scored with a human-readable na_reason, the composite renormalizes over the factors that did score, and the response reports scored_factors / total_factors. An instrument where too few factors score is graded NR (not rated) rather than assigned a fabricated number.

​
Calibration references

The model is sanity-checked against published third-party assessments of the same instruments — most directly Particula’s public AA+ rating of Anemoy’s JTRSY (the engine must land within ±1 notch on the same instrument), alongside Cicada’s tokenized-risk-assessment and structured-credit index reports. The reference reports are archived in the engineering repository under archive/credit_risk-reference/ (Particula/Anemoy JTRSY, May 2025; Cicada index dashboards and tokenized risk assessments). These are cross-reference points, not inputs: Alterscope does not resell or rebroadcast third-party ratings.

​
Access

Browse ratings in the Alterscope app on any plan. Programmatic access via API key requires the Enterprise plan — under-tier keys receive 403 ENTERPRISE_PLAN_REQUIRED with an upgrade link. The endpoint surface and response shapes are documented in the API reference and the RWA risk grades concept page.