Framework 01 · Policy

Horizon Policy.

Probabilistic forecasts on upcoming policy instruments — consultation papers, statutory instruments, grid-code revisions, contracts-for-difference design — across forty-seven jurisdictions. Calibrated against actual outcomes every quarter, published with provenance.

Request a demo Read the method →

What it forecasts

Five categories. Eighty per cent of write-downs.

Across the renewables capital stack, eighty per cent of investment write-downs trace back to five categories of policy change. Horizon Policy is built around those five.

Connection-queue reform

Gating regimes, milestone enforcement, queue cleansing, behind-the-meter rules. The single largest driver of write-downs on UK and Texas portfolios since 2023.

Contracts-for-difference design

Auction caps, indexation rules, technology lanes, eligibility windows. Small drafting changes shift IRRs by hundreds of basis points.

Planning rule changes

NSIPs, DCO regimes, distributed-generation consenting, BESS-specific consenting. The slow-motion variable that decides which sites survive.

Fast-track regimes

REPower-style accelerations, strategic projects of common interest, emergency consenting. High variance, high optionality value when forecast early.

Consenting timelines

Statutory consultation windows, examination periods, judicial review exposure. Variable by jurisdiction; forecastable from caseload backlogs and political calendars.

Public calibration

Current 24-month Brier score on grid-code revisions across UK, Germany, Ireland and Texas is 0.087. Calibration plots are published quarterly in the Library.

How it works

Corpus → event graph → calibrated forecast.

Step 1 · Corpus

41 million documents

Every consultation paper, statutory instrument, parliamentary committee transcript, judicial review filing, regulator letter, NESO publication, ENTSO-E and ERCOT operational document. Refreshed daily; archived with version history.

Step 2 · Event graph

Documents linked across years

A consultation paper in 2024 is rarely a surprise — it is the third predicted move on a chain we can trace back four years. The event graph encodes those chains and exposes them to the forecasting layer.

Step 3 · Calibration

Forecasts scored quarterly

Every closed forecast is scored. Brier, log-loss, reliability diagrams, all published. Drift is flagged within one quarter. Calibration class is tracked separately for each instrument category and jurisdiction.

Worked example · UK CfD AR7 In April 2024, Horizon Policy assigned a 0.61 probability to the UK government adopting an indexation reform inside the next CfD allocation round (90% interval [0.48, 0.73]). The reform was adopted in November 2024. The reasoning, document chain and probabilistic trajectory across the eight-month window is preserved in the audit log; clients can reproduce the chain through the Horizon Policy interface.

Sample outputs

What the IC memo looks like.

Instrument	Jurisdiction	P(this window)	90% interval	Calibration class
Grid-code revision — connection gating	UK / NESO	0.74	[0.62, 0.84]	UK grid-code · 318 closed
CfD AR8 indexation rule	UK / DESNZ	0.41	[0.28, 0.55]	UK CfD design · 14 closed
Onshore-wind planning fast-track	UK / DLUHC	0.58	[0.44, 0.71]	UK planning · 27 closed
EEG renewables remuneration revision	DE / BMWK	0.36	[0.22, 0.52]	DE EEG · 19 closed
ERCOT interconnection reform	US / Texas	0.68	[0.55, 0.80]	ERCOT IBR · 22 closed
EirGrid ECP-2.5 revision	IE / CRU	0.52	[0.39, 0.65]	IE connection · 9 closed

Illustrative outputs; live tenant figures will differ. "Closed" denotes the number of past forecasts in the calibration class whose outcome is now known. Wider intervals on thinly-evidenced classes are themselves information.

Who uses it

The desks Horizon Policy was built for.

Project finance teams — staging capex on policy probability triggers, locking optionality clauses in EPC contracts.
Renewables developers — walking away from sites whose viability depends on a rule with sub-35% forecast probability.
ESG asset managers — sizing sovereign and corporate exposure to upcoming carbon disclosure regimes.
Equipment OEMs — planning manufacturing footprint against jurisdictional eligibility windows.
Trade bodies and regulators — quantifying market-impact of draft instruments before publication.