Skip to content
LEXUNCreate your KeyRun a decision free
Country intelligence · Level 3 of 5

United Kingdom

This is the country page LEXUN can write honestly — and today it is the only one. Every figure below resolves to a named source: the bundled national datasets, or an outcome resolved through the public forecast register against a source named before the answer was known. Where the evidence stops, this page says so instead of changing the subject.

Latest recorded conditions

From the world state — derived only from bundled evidence, content-versioned, refreshed only by a release. Each figure is the latest LEXUN has recorded, with its period: official sources may have published newer figures since, and those appear here only once a release records them. Static copy below is written by the release process; the live view re-reads the same file in your browser.

Recorded indicators — world-state ws-04e04a9a2a (static copy, regenerated at every release)
IndicatorValuePublished / effectiveSourceOn the register
Bank of England Bank Rate3.75%2025-12-18 · confirmed 2026-09-17 by the registerBank of England001 WITHDRAWN · reason on record, never scored · 006 WITHDRAWN · reason on record, never scored · 007 RESOLVED TRUE · Brier 0.0529 · 008 WITHDRAWN · reason on record, never scored · 010 RESOLVED FALSE · Brier 0.2209
UK CPI annual inflation (12-month rate), as first published3.1%2026-09-16www.ons.gov.uk002 RESOLVED TRUE · Brier 0.0121 · 003 RESOLVED FALSE · Brier 0.25 · 011 RESOLVED TRUE · Brier 0.0144
UK headline unemployment rate, as first published4.9%2026-09-15www.ons.gov.uk004 RESOLVED TRUE · Brier 0.01 · 012 RESOLVED TRUE · Brier 0.01
Registered company insolvencies, England and Wales, latest resolved month1,9312026-08-18www.gov.uk005 WITHDRAWN · reason on record, never scored · 009 RESOLVED TRUE · Brier 0.0225

Figures are as first published; revisions do not change them here. A bundled snapshot can be stale between releases and says so rather than pretending to be live.

The four states of the world

Official past — what the record says happened. The table above: every figure resolves to a named official source, is stored as first published, and changes only when a release ships new evidence. Later revisions never overwrite it — what was believed at the time is itself part of the record.

Estimated present — deliberately absent. LEXUN runs zero live connectors, and the coverage registry is gated to say so. Between a stale snapshot that admits its date and a live-looking estimate nobody has validated, LEXUN ships the snapshot. A nowcast layer earns its place the way everything here does — by being scored against the official publication it tried to anticipate, in public — and none has yet.

Knowledge boundary — what LEXUN cannot see. The gaps are data too: the gap registry lists every known gap — no estimate of the present, no country beyond the UK, no city evidence, no research parameters, no calibration record, whole question classes refused, century-scale history not yet licensed — with why each exists, what it does to an answer, and what evidence closes it. Missing information is modelled as missing: it widens the uncertainty or refuses the question, and is never filled with confidence. The registry can never say “complete”.

Probabilistic future — many paths, never one. The open entries in the public register, each frozen with a sealed receipt before its outcome is knowable, and the simulations you run on the platform: thousands of paths with explicit uncertainty, not a single confident number. When reality arrives, the future state is scored against the official past — that loop is the product.

Keeping these four apart is the discipline: recorded fact, unvalidated estimate, admitted ignorance and simulated possibility are never blurred into one voice.

The historical record

Bank Rate, 1975–2025: the full official change history (258 observations, anchor-checked in the build) with time-in-band, episode and transition-count exhibits on the Accuracy Centre — counted frequencies with samples attached, explicitly refused as a transition matrix.

CPI yearly averages, 1989–2025: ONS series D7G7, bundled current-vintage-only and the file says so; the two-variable historical analogue panel shows which past years most resembled today — and why similarity is not repetition.

What the register has resolved here

Eight UK figures have entered the record as scored outcomes, and zero forecasts stand open. Every count on this page is read from the ledger at release time; misses stay on the record with the hits, and each entry has a permanent citable page.

No forecast stands open.

Four withdrawn with the reason kept: /f/001, /f/005, /f/006, /f/008.

Supported simulations

Three released models run for UK decisions: business cash runway, career-change buffer survival, and major-purchase buffer survival — each with its version, enforced 3–36-month horizon bound, refusal list and stated weaknesses in the model registry. Run unlimited decisions free; no account needed.

What is still missing — stated, not hidden

No city state. London evidence beyond the national series is not bundled, so no London-specific figures are shown — a city page will exist when city evidence does.

No approved research parameters. The research-to-simulation pipeline is future work; until it exists the UK sits at Level 3 on the coverage ladder, not Level 4.

No validated calibration record. Fewer than ten outcomes have resolved per domain, so no accuracy percentage is published for any model — the same sample rule the whole site obeys.

Why only one country? Because coverage is computed, never claimed: the registry holds the rule, and a second country page appears the day a second country earns Level 2.