Benchmark governance

Settlement-reference licensing

What licensing a benchmark to settle against requires — and why publication alone does not grant it. Reading a value is free; referencing it in something that settles is not the same act.

Status: published 2026-08-20 as oracle build item O0.6(c). No settlement-reference licence has been granted, and none can be while the independent committee seat is vacant — see §7. Applies to: the benchmark class — WETGRI, WETFED, WETX, WETFRAG, and any benchmark added later.


1 · The distinction this document exists for

A chart that is wrong is an erratum. A settlement reference that is wrong moves money.

Everything else in this governance set is calibrated to the first case. cessation.md gives thirty days' notice "so anyone referencing it can transition" — written with a journalist and a chart in mind. restatement.md corrects a value and publishes both versions, which is complete if the reader was quoting it and insufficient if they settled against it.

A settlement reference is a benchmark that a contract, instrument or product resolves against: the number determines who pays whom. That is a different relationship, it needs different commitments, and it is not created by publishing a value.


2 · Publication is not authorisation

Reading a W.E.T. benchmark is free. Referencing one in something that settles is not the same act, and is not licensed by publication.

Every benchmark value is served free and unauthenticated — every settlement row, every constituent list, the intraday indicative, and (once deployed) the on-chain feed. That is a standing commitment with a notice period, recorded at rule 6 of conflicts.md. Nothing here narrows it.

But a value being readable does not make it referenceable. Without a settlement-reference licence, a contract referencing a W.E.T. benchmark has:

  • no determination procedure — nobody is obliged to say what the final value is;
  • no disruption fallback — see §5, and note the benchmark will not invent one;
  • no restatement entitlement — see §6;
  • no cessation notice entitlement beyond the ordinary thirty days written for readers;
  • and no counterparty at W.E.T. There is nobody to call.

This is a statement of fact, not a sales position. We cannot stop anyone referencing a public number, and we do not claim the power to. We are saying what they will and will not have when they do.


3 · What a licence requires

Cumulative. All of them.

  1. A written agreement naming the specific benchmark, the specific instrument, and the settlement date or schedule. A blanket licence to "use W.E.T. benchmarks" is not granted: the obligations below are sized to a known exposure, and an unbounded one cannot be sized.
  2. The licensee's contract must specify its own fallback for a refusal, a stale feed and a cessation. See §5. We do not approve the fallback and we do not supply one. We require that one exists and that it does not resolve to a number we did not publish.
  3. The reference must be identified precisely: the benchmark, the settlement date, and whether it is the final or provisional record. A contract that says "the WETX level" without saying which record has not specified a reference — a settlement date can carry more than one row (§6).
  4. Attribution, as for every other surface.
  5. Disclosure to us of the approximate notional referencing the benchmark, updated annually. We do not publish it, and it is not a fee input. It exists because §7 and cessation.md require us to know whether a methodology decision has money behind it, and a governance body that does not know what depends on its decisions is not exercising judgement.
  6. The independent committee seat must be filled (§7).

What the licence does not include: any influence on methodology, constituent selection, weighting, or the settlement calendar. A licensee gets standing to be consulted under §4. They do not get standing to be accommodated.


4 · Determination — who says what the value is

The administrator determines the final value, applying the published methodology and nothing else, on the timetable below. The independent committee member is consulted on any determination that departs from the ordinary settlement path, with a minimum of five working days (committee-independence.md §4).

Ordinary determination The published settlement row for the date, once final — i.e. once the settlement date's UTC day has ended. Until then the row is provisional and may be replaced by a same-day re-run.
The provisional/final distinction is load-bearing A contract referencing a provisional record is referencing a value that can still change. This is not a defect and it is not hidden; it is why the record carries an explicit state field. Contracts should reference finals.
Timetable A determination request is answered within five working days. Where the answer is "the index refused on that date", it is answered in one.
Standard of decision The published methodology as it stood on the settlement date. Methodology changes are prospective (methodology-change.md); a determination never applies a later rule to an earlier date.
Disputes Through complaints.md, which answers challenges even when they are wrong and publishes the answer where the challenger is willing. A determination that is disputed is published with the dispute.

5 · The refusal case — the most consequential paragraph here

A refusal is not a zero, and it is not the prior value.

When a benchmark cannot meet its published floors it writes a row with status: insufficient_data and no value, carrying the class and rule that refused it. This is the methodology working, and it is not rare: WETX refused on five consecutive days, 2026-08-14 to 2026-08-18 (diagnosed in wetx-coverage-2026-08.md), and WETFRAG refused twice in the same fourteen-day window on a different rule.

On such a day:

  • W.E.T. publishes no value and will not supply one on request. Not zero, not the previous close, not an estimate, not an indicative. cessation.md states the reason directly: "yesterday's value is not carried forward. Filling a hole to keep a chart continuous is a fabricated print, and it is the single most tempting corrosive act available to an index." That refusal is inherited here exactly, and it is not negotiable for a licensee.
  • The refusal is published, signed, and machine-readable, carrying its class and rule so a consumer can distinguish undelivered books from a genuinely thin board. Those have different implications for whether tomorrow will print.
  • A refusal read as a value is the failure mode this whole programme is built against. The canonical checksum string encodes a refused value as 0.000000 — correct for a digest, and catastrophic if a consumer reads it as a price. WETFRAG trades near 1.0; a refusal read as a number is a 100% drawdown that never happened.

Therefore: a licensee's contract MUST specify its own refusal fallback, and we require it in writing before granting a licence. A contract that has no answer for "the index did not print today" has an undefined outcome on a day that has already occurred five times this month.

A flagged fallback is a third case. Where a methodology defines its own fallback — WETFED reverting to the nearest available meeting cycle — the value is published and flagged as a fallback. A licensee must decide in advance whether their contract accepts a flagged fallback as a settlement value, because it is a real number computed on a substitute basis, and treating it silently as an ordinary print is the exact fold cessation.md forbids.


6 · Restatement — what a licensee is owed, and what they are not

A published value may be corrected under restatement.md, for input error, computation error or settlement-source error — never because it looks wrong.

What happens mechanically: the corrective row is appended, the original stays flagged superseded, and the corrective row's previousChecksum is the superseded row's checksum — so the chain extends through the correction and a restated series still verifies end to end. A settlement date can therefore carry more than one row. A licensee must key on checksum, never on date.

What a licensee is owed:

  • Direct notice within one working day of the restatement being published — not merely a published row they might notice. This is the substantive difference between a licence and a free read.
  • A written statement of the error, its cause, and the affected dates.
  • The 90-day horizon applies to them too: errors confirmed more than 90 days after publication are disclosed and analysed but not restated, because beyond that horizon the value has been relied upon and stability is worth more than the correction. A licensee does not get a longer window than the public; they get faster notice within the same one.

What a licensee is NOT owed, stated plainly because the alternative is implying otherwise:

  • No unwind, no reversal, and no compensation. W.E.T. does not settle contracts, hold positions or clear anything. If a contract has settled against a value later restated, that is between the contract's counterparties, and their agreement must say what happens. On-chain settlement in particular is irreversible in a way a chart is not, and we will not represent otherwise: a restatement corrects the record, it does not undo a payment.
  • This is precisely why §3(2) requires the licensee's own contract to define its restatement handling before a licence is granted.

7 · Conflicts, and the one thing the administrator may not do

While a settlement reference is live, the administrator may not:

  • hold any position in the referencing instrument, or in the constituents (already rule 5 of conflicts.md, and it binds harder here);
  • accept any fee that varies with the referencing instrument's volume, notional or outcome. A licence fee is fixed and published; a fee that scales with what settles against the number gives the administrator an interest in the number;
  • take a material methodology decision affecting the referenced benchmark without the independent committee member, and without disclosing what notional was known to depend on it.

AND THE STANDING BLOCK. committee-independence.md §6 states it and it is repeated here because this is the document someone will read first:

No settlement-reference licence may be granted while the independent committee seat is vacant.

A benchmark that something settles against, governed by a single person who is also its commercial beneficiary, is the structure the conflict wall exists to prevent. The vacancy is a disclosed gap for a published number and a disqualification for a referenced one. The seat is vacant today.


8 · Cessation, and the window nobody plans for

cessation.md gives thirty days. For a licensee that is the start of the procedure, not the whole of it — the substantive treatment, including what happens between notice and stop, is in determination-and-cessation.md, which this document requires every licensee to have read.

The short version: a cessation notice does not suspend the benchmark. It keeps publishing, normally, for the notice period, and a contract settling inside that window settles against a real value. The failure to plan for is not the stop — it is the last thirty days.


9 · Applying

Through the contact in complaints.md. An application will be refused today, under §7, and the refusal will say so rather than being left to elapse. That is not a formality: an enquiry answered honestly with "not while the seat is vacant" is more useful than one held open.

The rest of the set
Methodology change & consultation
What counts as a material change, the 14-day consultation window, and why changes are prospective.
Data licence & rights
What the free tier grants, what attribution it requires, and what commercial use needs.
The loop architecture
The nine scheduled agents that cultivate indices, which one may publish, and the six mechanisms that assure accuracy.
The confirm gate
What human confirmation covers — merges only — and how to read the provisional and single-venue counts.
Restatement policy
The three narrow conditions under which a published value may be corrected — and the many under which it may not.
Cessation & fallback
What happens when a benchmark cannot be computed, and why the history stays published forever.
Conflicts & the conflict wall
The wall between index operations and the trading circuit, plus standing disclosures.
Data-error playbook
Detection, triage, quarantine — and the commitment to disclose errors nobody noticed.
Complaints & challenges
How to challenge a value, a rule, or a constituent. No account required.
Determination, disruption & cessation
Written for contracts rather than charts: who determines a final value, the disruption fallback ladder, and why a refusal is not a cessation.
The independent committee seat
The seven-part independence test, published before a candidate exists; the power to block rather than comment; and what the administrator may do while the seat is vacant.
Incident process
Who is paged, what they do, and what gets published when the machinery fails — as distinct from a wrong value, which is the data-error playbook.
Finding — the WETX coverage breach, August 2026
Why WETX refused for five consecutive days, the change that caused it, and what is still not established. Published because an availability figure without the incidents behind it is a number nobody can check.