PredictonomyPredictonomy
Home
Countries
Indicators
Sectors
Studios
Briefings
Methodology
Sign in
PredictonomyPredictonomy

Live macro and micro economic signals, historical context and probabilistic forecasts — built to surface trends before they become consensus.

Research and information only. Not investment advice. Forecasts are probabilistic and uncertain.

Explore

  • Home
  • Countries
  • Indicators
  • Sectors
  • Studios
  • Briefings
  • Methodology
  • Global dashboard
  • Signals

Company

  • Pricing
  • Developers
  • Data status
  • About
  • Legal
Methodology/Sector composites

Methodology · layer

Sector composites

Sector-health pages summarise many signals into one 0–100 composite. The UK mortgage-stress score is the first such composite; the same pattern generalises to other sectors.

Composite

UK mortgage-stress score (0–100)

A transparent weighted average of five component scores. Each component is itself the average of a change score, a level score and a trend score, clamped to 0–100. Series breaks and missing observations are excluded from the score so anomalies cannot inflate it.

Arrears pressure

Balance, share and possessions — rate of change and level (FCA/BoE MLAR, quarterly).

35%

Demand pressure

Approvals and net lending trend — falling demand raises stress (Bank of England, monthly).

20%

Rate pressure

New and outstanding mortgage rates and fixes — level and change (Bank of England, monthly).

20%

Price pressure

House-price MoM/YoY — falling prices raise pressure (ONS HPI / Land Registry UKHPI, monthly).

20%

Affordability proxy

Rates vs prices vs lending — a combined affordability trend score.

5%

Derived metrics

How movements are computed

  • Frequency-aware change: MoM/YoY for monthly series, QoQ/YoY for quarterly series.
  • Period-on-period changes to or from a zero/missing endpoint are suppressed (never shown as a −100% move); changes beyond a configurable threshold are flagged as a series break / unit change and shown under review rather than as a real move.
  • Moving averages (3m/6m for monthly, 4q for quarterly), 5-year z-score and 10-year percentile rank give historical context.
  • Revisions preserve the previous value and increment a revision number; freshness is tracked per series.
Research use only: This page explains research methodology only. It is not legal, regulatory, financial, investment, tax, mortgage, or property advice.