Every month the FEP engine's district ranking is frozen as an immutable prediction. When the lagged
outcome data publishes months later, each prediction is scored against what actually happened — ranking skill,
precision and calibration, hits and misses alike. This page watches that process live.
Loading…
Prediction ledger
Frozen FEP predictions awaiting or carrying an outcome score. Each is immutable once written.
What is tested here — and what is not
Honesty about scope: only a tool that makes a falsifiable prediction can be scored against outcomes. We do not claim to validate tools that are not forecasts.
Tool
Makes a prediction?
In the backtest?
FEP — Frailty Emergence Probability
Yes — a daily district-level forecast of emerging frailty
Under test
Winter Readiness Index
Yes — a forward seasonal forecast
Joins once a winter season closes
Rural Access Vulnerability (RAVI)
No — identifies current hotspots
Intelligence, not a forecast
FEP × Carer Burden Quadrant
No — overlays two current measures
Intelligence, not a forecast
Reachable Neighbourhoods
No — ranks current need for outreach
Intelligence, not a forecast
Community Touchpoints
No — maps existing community assets
Intelligence, not a forecast
NHS Pressure Intelligence Map
No — describes current system strain
Intelligence, not a forecast
Why only FEP
Most of the suite are intelligence instruments — they describe where need is now, which is useful but not a claim about the future that reality can later confirm or refute. FEP is different: it forecasts where frailty is emerging, so it can be frozen and scored against what actually happens. Winter Readiness is also a genuine forecast and will enter the backtest once a season completes and its outcomes publish. We score what can honestly be scored, and say so plainly about the rest.
Latest scorecard
How the most recently scored prediction performed against realised outcomes.
Data sources flowing in
Freshness of each open-data feed. "Fresh" means refreshed within its expected cadence.