Curation risk
On isolated-market protocols the protocol does not choose which collateral is acceptable or at what LLTV — a curator does. The curator picks the markets, sets the caps and can re-point a vault's allocation. Their judgement is the risk.
There is no objective measure of judgement, so curation is scored on proxies that are at least verifiable: how much capital has been entrusted to them, whether their identity is attested, and how the role was granted.
Curator score
Both supported curator registries produce a 1-5 score, lower is better.
Assets under management sets the base tier. It is the market's own verdict: capital is the one signal that cannot be self-declared.
| AUM | Tier |
|---|---|
| $100M or more | 1 |
| $10M or more | 2 |
| $1M or more | 3 |
| $100k or more | 4 |
| below $100k, or unknown | 5 |
Adjustments, added to the tier and clamped to 1-5:
| Registry | Adjustment |
|---|---|
| Morpho | +2 when the curator is not verified |
| Euler | +1 when the curator has no registry name, +1 when the address holds the vault only as owner rather than an explicit on-chain curator role |
Euler's registry has no verification flag, so an entity-registry name stands in for it — and only a registry name counts. A name inferred from a vault title or an explorer tag is a display label, not an identity attestation, and does not earn the bonus.
The two adjustments are different sizes because the registries differ: Morpho's
verified is an explicit attestation whose absence is meaningful, while Euler's
missing name is more often incomplete coverage than a red flag.
Reputation allowlist
An allowlist of established professional risk managers caps a matched curator at score 2. It is matched case-insensitively against the curator name, substring aware.
This exists because AUM is a lagging indicator: a well-known risk manager launching a new vault starts at zero AUM and would otherwise score 5 on their first day. It only ever improves a score, never worsens one, and it is a curated list with a stated purpose rather than a scoring input anyone can game.
Market-level curation
For a curated market, the number of curators is itself a signal:
| Curators on the market | Level |
|---|---|
| 2 or more | green |
| exactly 1 | yellow |
| none | red |
The market's curator score is the allocation-weighted average of its curators' scores where allocations are known, otherwise an equal-weight average. A market with no curator at all scores 5 at the protocol level; when it appears inside a vault it is capped at 3, because unvetted is not the same as proven bad.
The curator score is then combined worst-of with the market's bad-debt rating — see Bad debt. Observed loss beats predicted quality: a market with two blue-chip curators and 70% bad debt is rated on the bad debt.
Limits of this dimension
- AUM measures trust already given, which is partly reflexive.
- Verification confirms identity, not competence.
- A curator repeatedly associated with bad-debt markets is not yet penalised on their own score; the penalty lands on the affected markets and on any vault holding them.
- Curator identity on some protocols resolves to an address with no name at all. Those are scored on AUM and role, flagged as unrecognised, and capped at yellow in vault ratings rather than assumed bad.