Skip to main content

Weather Risk Stress Index

A classifier of weather-forecast instability and weather relevance.


The Weather Risk Stress Index (WRSI) classifies how unstable weather forecasts are (ensemble spread, run-to-run change, model disagreement) and whether weather matters for demand at this time of year.


WRSI answers "can I trust the current weather forecasts, and does weather matter right now?". It is a classifier, not a forecast, and carries no forward, predictive or sizing claim.

Overview​

WRSI focuses on identifying and tracking persistent weather regimes that shape baseline assumptions for energy demand and system behaviour.


It addresses questions such as:


  • How unstable are the current forecasts?
  • Do the main weather models agree?
  • Does weather matter for demand at this time of year?

WRSI is not a forecast. It describes forecast conditions, not weather or price outcomes.

Interpretation Guide​

WRSI is designed to provide context around prevailing weather regimes and their implications for energy demand and system behaviour. It should be interpreted alongside forecasts, realised weather data, and seasonal norms.

Reading elevated instability​

Elevated WRSI readings indicate that forecasts are moving between runs, disagree across models, or have wide ensemble spread. Read this as lower confidence in any single forecast run.

This does not imply a demand or price outcome.

Reading regime transitions​

Shifts between regimes mark changes in forecast confidence, often near seasonal inflection points when weather's influence on demand changes.

Relationship to events​

Where applicable, WRSI Events may highlight notable regime changes or breaks from established patterns. These events provide context around moments when weather assumptions may need reassessment.

Use within a broader framework​

WRSI is most effective when used as a structural input alongside short-term forecasts, realised weather data, and demand-sensitive market analysis.

Data notes (October 2026)​

  • skill_score is a ratio, not a 0-100 score. It is 1 - MAE / climatology MAE: above 0 beats climatology, below 0 is worse. Typical values are negative (mean about -1.19 for GEFS and -0.31 for ECMWF); read it as a relative monitor.
  • Climatology is point-in-time. It uses the same calendar day in earlier years only (minimum three years). It previously used later years and the verifying actual. avg_skill_score and confidence_score moved; confidence_label, wrsi_score, wrsi_regime, events and the backtest are unchanged.
  • Earlier fixes (30 September 2026): GEFS same-cycle join and ECMWF de-duplication, and ECMWF agreement extended back to March 2024.

Explore WRSI further​

To explore WRSI in more detail: