Data Methodology

This page documents the trade data engine used to generate the Trade Signals layer. All indicators are derived from build-time Eurostat COMEXT snapshots and computed deterministically (no runtime API calls).

Definitions

COMEXT: Eurostat’s trade statistics system used as the source layer for this site.

Dataset: one JSON snapshot representing a fixed combination of reporter (EU/FI), partner group (CN/ASEAN6/US/JP), SITC section (0–9), and monthly time range (~24 months).

Reporter: EU aggregate or Finland (FI), depending on dataset.

FLOW = 1: imports (the current trade engine is import-focused).

Values: nominal EUR values as reported by COMEXT (no inflation adjustment; no seasonal adjustment).

SITC 0–9 explained

The trade engine groups products using SITC Rev.4 sections (0–9). These are broad categories intended for macro-structure analysis rather than product-level supplier inference.

SITC 0Food and live animals

Agricultural and food supply chains; seasonal patterns and policy sensitivity can dominate signals.

SITC 1Beverages and tobacco

Branded consumer goods with excise taxes and regulatory constraints; often stable but policy-affected.

SITC 2Crude materials, inedible (excl. fuels)

Upstream inputs such as timber, pulp, ores, rubber; typically commodity-like pricing and cyclical demand.

SITC 3Mineral fuels, lubricants, related materials

Oil, gas, coal-related products; geopolitics and price shocks often outweigh gradual trade structure change.

SITC 4Animal & vegetable oils, fats, waxes

Edible oils and industrial fats; tied to agricultural output and food processing capacity.

SITC 5Chemicals and related products

Industrial chemicals and related products; higher value density and compliance exposure (standards, regulation).

SITC 6Manufactured goods (chiefly by material)

Material-based manufactures (metals, paper, textiles, wood manufactures); often foundational to industrial supply.

SITC 7Machinery and transport equipment

Machines, electronics, vehicles/parts; higher complexity, longer lead times, and stronger supply-chain constraints.

SITC 8Miscellaneous manufactured articles

Broad bucket for finished goods (household items, tools, consumer articles); commonly relevant to durable goods.

SITC 9Commodities and transactions n.e.s.

Residual/special transactions category; interpret cautiously and avoid over-weighting in strategic conclusions.

Practical note: SITC 6–8 tend to be the most relevant for durable goods and industrial procurement signals.

Data source

Eurostat COMEXT (dataset: ds-059331). Monthly import data, limited to the most recent ~24 months. Data is ingested at build-time and stored locally under src/data/comext/. No runtime API calls are made.

Architecture

  • Build-time ingestion only
  • Canonical partner_mix dataset structure
  • Geo-split reporting (e.g., EU vs FI)
  • Deterministic JSON snapshot outputs
  • Manifest-based dataset indexing (index.json)

Baseline definition

Shares are calculated within a restricted partner subset: China (CN), ASEAN-6 (TH, VN, MY, ID, PH, SG), United States (US), and Japan (JP). Shares represent proportions within this defined baseline — not total global trade share.

Signal definitions

Share: partner_value / baseline_total

24-month share delta: latest_share − share_24_months_ago

PDI_max: top_share × (1 + top_share_delta_24m)

PDI_CN: cn_share × (1 + cn_share_delta_24m)

12m vs 12m growth: rolling 12-month aggregate compared to the prior 12-month aggregate.

Interpretation note: PDI terms are designed to highlight concentration risk and its recent movement (within the restricted baseline). A higher value indicates a stronger “dominant partner + momentum” signal.

Data limitations

  • Monthly data may be revised
  • Nominal EUR values (no inflation adjustment)
  • No seasonal adjustment applied
  • 24-month window limits long-cycle inference
  • Restricted baseline excludes other global partners
  • SITC sections are broad; they support structure signals, not product-level sourcing conclusions.