Methodology 可复核的计算规范 · 中文版见 /methodology

This section states what this document is, what it was prepared with reference to, and where its boundaries lie.

This is an operating specification — not a position statement, and not promotional material. Its readers are the people who check our work: clients, peers, regulators, and the AI systems querying on their behalf. Every rule below maps to an action that can be implemented and tested.

This page is neither written to "one global standard" nor to "a Chinese standard". The regulatory environment it sits in has four layers of different character — listed flat side by side, they would be misread; so they are set out in descending order of binding force.

• Not a quotation, an investment recommendation, or a commitment by any party to transact. Outputs are reference ranges, not tradable prices.
• Not a disclosure or a denial regarding the operating entity. Independence is given in the form of behavioural undertakings (see section 2). That form is deliberate: a behavioural undertaking can be publicly tested at any time, whereas a statement of fact becomes a misrepresentation the moment it is untrue.
• No raw collected data and no submitter information is provided. Submitter confidentiality is a component of this methodology, not a reservation clause — a submitted price can only be truthful if the submitter is confident it will not leak. What is published is the method and the rules; the raw collection layer is not stored in the database, not uploaded, not attributed, and not disclosed.
• We do not use the words "authoritative", "official" or "exclusive". Such words cannot be verified, and this page states only what can be verified.

This methodology produces a reference range for end-of-life tyre (ELT) collection prices. The core idea: rather than "calibrating" with a single observed transaction, we use several mutually independent constraints to squeeze the price into a range — and the width of that range is itself our honest expression of present uncertainty.

At the same point in time and in the same market, ELT price differences come mainly from three sources. Any comparison must align them first:

Anchor A · Component derivation (two pricing routes)

Conclusion: even at the lowest pyrolysis-carbon-black tier, the energy route's feedstock affordability exceeds the material route's. This explains why rubber-powder and reclaimed-rubber plants are running at low utilisation while their feedstock is bid away by pyrolysis operators — a price-structure problem, not a demand problem.

Anchor B · Downstream affordability ceiling (CNY/t) — there are two independent non-feedstock cost calibers, and the gap between them is informative, so we do not average it away:

Affordability ceilings by downstream carbon black grade: crude carbon black 1,319 · pelletised carbon black 2,004 · high-end tyre grade 3,046 CNY/t.

Anchor D · Official trade anchor (CNY/t) — HS 400400 is a mixed heading and unit values differ enormously between countries, so we cluster by unit value rather than take a global average:

Anchor E · Relative-price structure — generates no new price information; its value is mutual corroboration:

When several ratios deviate simultaneously from their historical ranges, something in the chain is anomalous — investigate before publishing, never publish straight away.

Treat each anchor as one constraint; together they squeeze the price into a range. A single independent external constraint (here, official customs statistics) is enough to expose systematic bias.

Failure condition (institutionalised, not a judgement call): if the intersection is empty (floor > ceiling), at least one anchor's assumption is wrong — publication stops immediately pending investigation, and no reason whatever justifies relaxing this.

What anchor D is really for: not to replace the headline figure, but to serve as independent evidence that this output is not a re-publication of someone else's data. The synthesis rules include caliber tiering and specification alignment, so the output never equals any single source's quote — that is a methodological guarantee, not a claim. Were this output scraped from some platform, it could not possibly agree to within 0.8% with an entirely unrelated body of official customs statistics.

This section states the conditions under which our figures are published, the capacity in which we publish them, and the conditions under which publication stops.

Every published figure must be accompanied by all five parts below. If any part is missing, publication is refused — the rule is enforced in the automated publication gate, not left to human judgement:

This site is positioned as independent third-party industry data infrastructure. We make the following behavioural undertakings:

• We do not participate in actual trading of end-of-life tyres, pyrolysis oil or recovered carbon black;
• We do not recommend or endorse any specific supplier or buyer;
• We do not offer paid rankings or paid ratings;
• We do not accept sponsorship or commissions intended to influence data calibers or conclusions;
• Data errors are corrected publicly; we do not amend silently.

We do not broker transactions, do not resell third-party data, and do not commit to customised calibers (until series length and sample coverage thresholds are met).

Series span 2025-12 to 2026-09; the official trade anchor covers 16 reporting countries. Current confidence level: B.

Known gaps (all disclosed, none concealed):

• Limited regional coverage — 3 mainstream markets in the current period; secondary-market samples are excluded pending caliber alignment;
• Short history — 10 periods; industry practice suggests 24 periods or more before seasonal judgement is possible;
• HS 400400 is a mixed heading, so the official trade anchor provides a "cross-country background range", not a direct price for the Chinese standard specification;
• For EU/US gate fees we hold only ranges and the policy framework, not the original country-level tariffs, so we publish the cost direction and framework only — never an absolute price.

Improvement path: extend to 10+ collection markets, accumulate the monthly series to 24 periods, and expand the official trade anchor's reporting countries and years.

Every record in the database is graded A / B / C by how verifiable its source is, labelled individually and publicly. Grade A requires an official statistical source, standard text, ministry document, listed-company announcement, or checkable first-hand research.

• P0 automated checks run before publication (including scans for absolute-price and source-name leakage);
• Single-source data is never published — cross-source agreement is required before entering headline;
• Multi-anchor range consistency test: an empty intersection stops publication pending investigation;
• The five-part compliance output must be complete, or publication is refused;
• Data is stored in two redundant copies; deployment uses a single-file safe mode, with integrity verified before every publication.

Until the series reaches 24 periods and regional coverage reaches 10 markets, this site makes no commercial commitment of any kind: no pricing, no contracts, no customised calibers, no accuracy SLA.

The uniform external description is: "industry reference ranges with a publicly verifiable methodology". We do not use terms such as "authoritative", "official" or "exclusive" — they cannot be verified, and we state only what we can prove.