The UK economy,
in perspective.

Higher = stronger activity relative to its recent historical pace. Equal-weight average of component z-scores. Each component uses its own preceding 40 calendar quarters, with at least 12 observations, excluding the current quarter. Extended history before the first full basket uses at least half the components (minimum two), with weights divided equally among those available. Coverage changes can alter the score and limit comparisons across eras. Once the full basket starts, incomplete baskets are omitted; current scores require every component.

Quarterly releases retain their reported values. Monthly observations are averaged only when all three months are available. Daily observations are averaged within completed quarters, requiring at least 30 observations and a value within seven days of quarter end. UK gas uses the mean of available weekly observations. Growth compares adjacent quarters; annual changes compare four quarters earlier. Real wage growth = wage-level ratio divided by (1 + CPI inflation / 100), minus one. Labour inputs already cover rolling three-month periods; the quarter average smooths overlapping estimates.

Extended history uses the available components before the full basket starts, with at least half the inputs and a minimum of two. Missing data are never filled or invented. Full-basket-only history is available using the checkbox above the cards. It remains unavailable until all sources pass the site's freshness checks. The latest complete model quarter may lag current releases. Source checks follow the existing daily refresh while the site is used; scores change when sufficient new data arrive. These are custom descriptive models using revised histories, not forecasts, official indices or recession probabilities. No predictive performance is claimed.

Underlying source links and release dates are available on the corresponding indicator cards. Shorter model histories reflect the availability of all required inputs and the initial calibration period.

100 × expanding sectors ÷ 4, equally weighted: agriculture, production, construction and services. Expansion means strictly positive quarter-on-quarter real output growth; zero counts as unchanged. These are four broad, non-overlapping sectors, not a detailed industry diffusion index or GDP-weighted contribution. Production is already a growth series; other sector indices are converted to growth. All four are required.

Calculated from the existing source feeds, checked daily while the site is used. Missing observations are not fabricated. Current readings require all inputs to pass freshness checks. Historical observations may be revised.

Experimental ridge regression (fixed penalty 10), using production and retail 3-month growth, unemployment change and last available quarterly GDP growth. At least 32 earlier GDP outcomes are required; the latest 60 train each estimate. Scaling uses only its training data. Historical estimates use the same elapsed day of the quarter as today and assume every monthly input and GDP release becomes available 60 days after its reference period ends. Actual historical release timestamps and original vintages are unavailable: this is a lag-assumed reconstruction, not a real-time backtest. The early-quarter estimate may be based entirely on previous-quarter activity. The uncertainty range uses the 90th-percentile absolute error from up to 40 earlier scored estimates (minimum 20), available by each estimate date. It is an empirical error range, not a guaranteed 90% probability interval. Pandemic quarters are retained. Current data revisions can change the reconstruction.

Calculated from the existing source feeds, checked daily while the site is used. Missing observations are not fabricated. Current readings require all inputs to pass freshness checks. Historical observations may be revised.

What GDP nowcasting means. This is an independent experimental model, not the Bank’s model or forecast.