Introduction
Cleaton answers one question about a pool, in a form both humans and contracts can consume: how long can its liquidity be expected to hold, and how confident is that estimate?
Argued in full in the whitepaper at §1.
Total value locked tells you how much is in a pool right now. It tells you nothing about whether that depth will still be there when you need to exit through it — and that is the property most allocation decisions actually turn on.
The measurement gap#
Two pools, both reporting $500M:
| Pool A | Pool B | |
|---|---|---|
| How it got there | Eighteen months of unsubsidised deposits | $17M → $500M in five weeks on a $29,000/day campaign |
| Median tenure | Nine months | Nineteen days |
| Concentration | Largest holder 4% | Top three hold 61% |
| Campaign expiry | None | Eleven days |
A human analyst separates these in ten minutes with a browser. An autonomous agent cannot, because the distinguishing information is not in any state it reads. As allocation moves to systems that consume only structured state, that gap stops being an analytics inconvenience.
What you get#
One call, four values, and a signature over all of them.
getAttestation(pool) → (horizon, confidence, expiry, signature)| Field | Meaning |
|---|---|
| horizon | Days the pool's liquidity is expected to stay above the retention threshold |
| confidence | Calibrated coverage — 0.7 means 0.7 against the published record, not a model's opinion |
| expiry | When the attestation itself goes stale. Distinct from the horizon |
| signature | EIP-712, recoverable to a registered attester |
The claim is falsifiable, time-bounded, cryptographically attributable and economically backed. Everything else Cleaton exposes is a different way of branching on those four values.
What Cleaton is not#
Not a rating. A short horizon is a statement about duration under current incentives, not a judgement about a protocol, its team, its code or its governance.
Not a yield forecast. The horizon is about liquidity persistence. Price and return prediction is a different and much harder problem, deliberately out of scope.
Not custodial. Cleaton never moves funds and holds no position in anything it scores. A durability oracle that also allocates has an interest in its own outputs.
Where to go next#
If you want to see a response, start with the quickstart. If you want to know whether the number can be trusted, start with calibrated confidence — that is the page that explains why the confidence is not simply the model’s own probability.