Stablecoin Risk Methodology
How we measure asset integrity for each token
Risk Tiers
Each token receives an integer score from 0–100 based on weighted risk signals. The score maps to a tier by inclusive floors — a score of exactly 25 is already Watch — and then passes the tier gates described below, which can demote or force a tier:
How the Monitor Runs
The facts below describe the production engine as deployed, not a design target. They are mirrored from the engine's own constants and checked in the same release as any retune.
Every tracked token is re-scored every 20 minutes. Each reading is timestamped and kept as history.
17 weighted components per reading (16 in the base weight table plus reserve-attestation staleness, which is always included). Price deviation from peg is the single heaviest — about a third of the total; two leading on-chain signals — anomalous minting and concentrated redemption drain — fire before price moves.
A signal with no data this reading is skipped and the remaining weights are rescaled to sum to one — so a missing signal's weight lands on the survivors. The per-token decomposition shows the effective weight that applied.
If a reading has no price data the last good score is carried forward for that cycle instead of being rescored on nothing. A price that stays pinned to the same value across readings (a dead feed) is frozen and flagged rather than read as a perfectly stable peg.
Risk Rating
The headline signal on every token is a letter grade from A+ to F. It’s mapped from the composite risk score (0–100, where 0 is safest and 100 is riskiest) — the lower the score, the better the grade:
The grade reflects a composite of peg stability, market stress, and structural risk (mechanism, governance, audits) — so it captures more than the live peg: a structurally weaker token won’t grade top even while on peg, while a token actively trading off peg is floored to a high risk score and grades accordingly. Bands are tighter at the safe end, where small score differences matter most.
What We Measure
The score is a weighted blend of risk signals. Price deviation from peg carries the single largest weight (about a third of the total); the rest is spread across the families below. Per-signal weights and the exact normalisation are published in the API docs and shown per token in the score decomposition. We weigh these signal families:
How far the current price sits from the token’s expected peg.
How long the price has stayed away from peg, not just an instantaneous blip.
How much it would cost to exit a meaningful position — thin liquidity raises risk.
Sudden spikes in price volatility relative to the token’s recent baseline.
Whether the price agrees across trading venues and across chains; disagreement is a warning sign.
Unusual supply changes and the freshness of reserve / collateral attestations.
Whether reference price feeds are fresh and consistent with on-chain trading.
Token Type Adjustments
Different token types have different expected behavior. The model adjusts so normal behavior isn’t mistaken for a depeg:
Traditional fiat-pegged stablecoins. Any meaningful move away from peg is penalized.
Yield-bearing tokens (e.g. sUSDe, sfrxUSD) are expected to trade above peg as yield accrues, so that premium is not treated as a depeg.
T-Bill backed tokens (e.g. BUIDL, mTBILL) whose value rises slowly with yields, so modest appreciation is expected.
XAU-pegged tokens (e.g. PAXG, XAUT) are scored against gold spot within a normal market spread.
Tier Gates
The tier is the score floor above, then four gates — so a published score inside a band can legitimately carry the tier below it, and a reading can be forced up:
- Low coverage — fewer than 5 signals carrying data while the price is at peg demotes a Warning to Watch (thin data inflates the renormalised weights).
- Confluence — a Warning needs at least 2 independent signals reading ≥ 0.1 (normalised) or it is demoted to Watch; Critical and a multi-source far-from-peg depeg are exempt.
- Severe depeg — a price at or below 25% under peg forces Critical and floors the published score at 80, whatever the other signals say.
- Premium — an OK token trading ≥ 2% above peg (yield, RWA and gold tokens excepted) is shown as Premium, an informational state, not a risk tier.
There is no consecutive-reading requirement: each 20-minute reading publishes its own tier.
Which score governs the grade
Two constructions run side by side for every token: the v3 framework composite(seven weighted categories graded at criterion level — the “Why this grade?” list on the token page) and the upstream pipeline risk score (the 0–100 score from the scoring engine). The published letter is always the worse of the two. When the upstream score is the one that governs, the framework categories explain the framework view, not the governing number — the token page says so where the grade is shown. Hard floors (sanctions designations, blocking flags) cap the grade regardless of either score.
Data Sources
Volume-weighted on-chain pricing and trading activity.
Estimated cost to exit a position at meaningful size.
Reference prices and feed freshness checks.
Independent pricing used to sanity-check on-chain quotes.
Want this data via API?
Every score, grade and signal on this page is available through the Webacy API — plug it straight into your own product.
Methodology identifiers: framework — · letter-grade scheme — · API schema —. This page was last reviewed against the production engine on 2026-08-19. Weights and constants are published in the API docs.
