How we source, score, and update

Our FOB Shanghai prices (6063-T5, 6061-T6, 6005A-T5, ACP 4mm PVDF) are derived weekly composites, not a private mill survey. Each price is built from three public benchmarks plus a typical export-to-spot margin assumption.

Published series: aluminum futures (Yahoo Finance ALI=F) in USD/t, SHFE main contract (post-close 15:00 CST), and Changjiang A00 spot (ccmn.cn 10:30 CST). FOB China is computed daily as aluminum futures (ALI=F) plus a fixed extrusion premium; architectural alloys and ACP use separate fixed premiums.

This methodology is intentionally fully reproducible from public sources; the formula is in scripts/scraper/tasks/ and changes are versioned by git history. Earlier copy implied a private "13-mill panel" — that was inaccurate language and was corrected in the 2026-04-22 content audit.

Landed cost = FOB price + ocean freight + insurance + destination port fees + import duty + anti-dumping + VAT + inland trucking. Each line is itemized in the calculator output so you can see what's driving the total.

• FOB — weekly composite from our price benchmarks (see §2 above), or your own input.
• Ocean freight — Freightos Baltic Index (FBX) origin→destination route, refreshed daily.
• Insurance — 0.3% of CIF default (institutional buyer rate); user-overridable.
• Duty — destination MFN rate from our tariffs table (HS chapter 76 lines), refreshed monthly from official tariff schedules.
• Anti-dumping (US/EU/CA) — order-specific ceiling rates from access.trade.gov / EUR-Lex; per-exporter rates may differ.
• VAT / GST — destination country rate applied on the duty-paid CIF base.
• Port fees + trucking — destination-specific defaults from forwarder quotes (Q1 2026 sample); user-overridable.

All currency conversions use PBOC / central-bank midpoints (see §1 sources). The calculator is a planning tool — actual landed cost depends on your forwarder, broker, and the specific HS line. Always confirm with a licensed customs broker before committing to a shipment.

TrustScore = clamp(
  SAMR_base (10 points if active registration)
  + customs_volume_score (0-30; ln-scaled against peer median)
  + certifications_depth (0-15; ISO9001=3, ISO14001=3, Qualicoat=3, IATF=3, GB/T 5237=3)
  + export_share (0-15; linear 0-100%)
  + market_fit (0-15; weighted by matches to requested buyer market)
  + listed_bonus (+5 if public-listed)
  - risk_flags (-25 max; UFLPA-adjacent, trademark-counterfeiting, AD evasion)
  - stale_filing_penalty (-5 if no customs filing in 90 days),
  0, 100
)

Tier A = 80-100 · Tier B = 60-79 · Tier C = <60. Tier labels shown on supplier cards and ranking tables. Raw per-mill component breakdown is downloadable from each supplier's profile page once the Supabase layer is live.

Current data status. This is the full scoring model. The customs-derived components (customs_volume_score, export_share, stale_filing_penalty) and certifications_depth are part of our roadmap and are not yet live — the customs and certification feeds are not yet available, so today those components score zero rather than being estimated. Current rankings score on registration identity, listing status, mill capacity, export-market breadth, and buyer-market fit.

Each ranking publishes its own abridged methodology in the body of the ranking page. In general: flagship rankings (annual) use a broader 6-factor weighting; market-specific rankings use destination-market export evidence as the primary signal (customs-volume weighting will apply once that pipeline is live).

No fraud case is published without at least two of:

1. Public dispute filing (court, arbitration, customs) or press coverage.
2. Submitted MTC, SGS, or spectro evidence from the affected buyer.
3. Customs record corroboration (volume drop, HS code mismatch).

The named seller is contacted for comment before publishing where possible. Their response (or refusal) is appended to the case record.

We fix errors. Publicly. With a note.

• Factual corrections (price, capacity, registration status) are published inline on the affected page with a visible note and the date we fixed it. An archived corrections index is on the roadmap.
• Editorial corrections (mischaracterization, unclear attribution) trigger an update to the article with a visible correction note at top.
• Take-down requests for defamation or false-fact claims are evaluated within 48 hours.
• 2026-07-23 — Privacy and cookie disclosures updated for consent-gated Google Analytics 4; advertising features remain disabled.
• 2026-09-02 — Restored consent-gated Google Analytics 4 after the 2026-08-22 load-on-visit change; GA4 runs only after you accept analytics cookies and can be withdrawn from Cookie settings in the footer. Advertising features remain disabled.