Skip to content
LEXUNGet your KeyRun a simulation
Coverage

One country. Stated plainly, enforced by the build.

Forecasting products love the word “global”. LEXUN’s coverage is a computed property of the evidence in the bundle — published at /coverage-registry.json, cross-checked by a release gate, and never typed by hand. Today that computation says: the United Kingdom at Level 3, and every other country honestly at Level 0.

The ladder

LevelNameWhat it requires
0UnsupportedNo verified evidence. Models refuse local reference figures; results state the gap instead of borrowing another country’s. Geography is still recorded and currencies display correctly — honesty about absence, not absence of the user.
1Global evidence onlyEvidence from global institutions, not specific to the country. LEXUN currently bundles no global-institution datasets, so no country sits here yet.
2National dataAt least one national statistical dataset bundled with provenance, licence and vintage discipline.
3National data + official reportsNational datasets plus figures resolved from official publications through the public forecast register — scored against sources named in advance.
4Local + researchCity or regional evidence and approved empirical research parameters. Requires the parameter registry, which is not built — so no country can claim this yet.
5Validated, backtested modelsCountry-specific models with a calibration record above the ten-resolved-outcome minimum. No country qualifies, and the accuracy pages say so in the same words.

The United Kingdom — Level 3 · country intelligence page

National data: the full Bank Rate change history (Bank of England, 258 observations, anchor-checked in the build) and ONS CPI annual averages (37 years, current-vintage-only and the file says so), both bundled with provenance and licence in the file itself.

Official figures through the register: CPI, unemployment and company insolvencies entered the world state as resolved forecast outcomes — each scored against the official source named before the outcome was known, using the figure as first published.

Why not Level 4 or 5: no approved empirical research parameters exist (the research-to-simulation pipeline is future work, recorded as such), and the resolved-outcome sample is below the ten-per-domain minimum every accuracy claim waits for.

Everywhere else — Level 0, and what that means in the product

A decision placed in Dubai, Istanbul or New York runs with the person’s own figures in their own currency — that arithmetic is universal. What LEXUN refuses to do is dress it in local authority it does not have: no UK base rate is presented beside a non-UK decision, no local survival prior is invented, and the result names the missing evidence as an open blind spot. A Dubai user getting a UK model with the currency symbol swapped is exactly the fabrication this page exists to rule out.

How coverage grows

A country moves up this ladder the same way the UK did: a verified national dataset with clean licensing enters the bundle (Level 2), official figures start resolving through the public register (Level 3), and — further out — validated research parameters and a real calibration record (Levels 4–5). Connector infrastructure for global sources is on the roadmap as future work; the registry will say so the day any of it is real, and not before. The claim “N countries supported” can only ever be printed from that file.

Machine-readable: coverage-registry.json · world-state.json · gap-registry.json (the known gaps, as data) · both regenerated by the release process and drift-gated. See the citation page for every endpoint.