Methodology
Risk Horizon synthesises publicly available regulatory output into structured intelligence using AI. This page explains the methodology, scope, data sources, and important limitations of that process.
Intelligence Generation
Risk Horizon runs a daily automated scan that processes publicly available regulatory publications — speeches, enforcement actions, consultation papers, supervisory guidance, and thematic reviews — across 18 jurisdictions. A large language model (Claude by Anthropic) reads and interprets this material, extracting structured intelligence signals with identified impact direction, materiality assessment, time horizon, and required action.
Signals are then grouped into risk themes, linked to scenario packs, and mapped to regulatory alignment frameworks. The entire chain from source publication to structured pack is AI-generated. No human analyst has reviewed individual outputs unless otherwise stated.
Daily scan cadence
Sources are scanned every morning at 06:00 UTC and processed within the same cycle.
AI classification
Claude categorises each signal by risk type, jurisdiction, materiality, and time horizon.
Scenario synthesis
Related signals are combined into scenario packs with trigger conditions and control stress points.
Regulatory mapping
Signals and scenarios are mapped to regulatory frameworks with clause-level relevance narratives.
Field Definitions
Confidence
AI-assessed certainty that the signal represents a genuine emerging risk. High: clear regulatory intent or enforcement action. Medium: indicative language or early-stage consultation. Low: background noise or speculative signals.
Materiality (1–10)
Estimated potential impact on a typical regulated financial institution. 1–3: monitoring only. 4–6: warrants management attention. 7–9: requires active remediation or escalation. 10: systemic or immediate board-level concern.
Impact Direction
Whether the risk represented by this signal is increasing (regulatory pressure or incident trajectory rising), stable (steady-state supervisory focus), or decreasing (resolved or de-prioritised by regulators).
Time Horizon
How quickly this signal is expected to crystallise into regulatory action or risk event. Immediate: 0–3 months. Near Term: 3–12 months. Medium Term: 1–3 years. Long Term: 3+ years.
Verification Status
Whether the signal has been cross-checked against a primary source. Verified means a source URL was retrieved and the underlying document was parseable. Unverified means the signal was inferred from context or secondary reporting.
Emerging
A risk cluster first appearing in regulatory commentary or supervisory focus. Not yet the subject of formal guidance or enforcement. Requires monitoring.
Crystallising
The risk is actively being regulated. Formal consultations, thematic reviews, or enforcement actions are in train. Immediate attention required by risk functions.
Established
The risk has been formally regulated and embedded in supervisory frameworks. Compliance is table stakes; focus shifts to evidence and ongoing controls.
Declining
Regulatory focus on this risk is reducing — either because it has been resolved, superseded, or de-prioritised in favour of newer themes.
Jurisdictions
Risk Horizon monitors publications from the following 18 regulatory bodies and standard-setting organisations. Coverage is limited to publicly available documents. Bilateral supervisory correspondence (e.g. individual firm SREP letters) is not accessible.
Risk Taxonomy
Every signal and theme is classified into one of twelve risk categories drawn from the Basel framework and standard risk taxonomy used across major jurisdictions.
Operational Risk
Risks arising from failed processes, systems, people, or external events, including business continuity and third-party dependencies.
Technology & Digital
Risks from technology failures, digital transformation initiatives, model risk, and IT infrastructure concentration.
Third-Party Risk
Risks introduced by outsourced service providers, supply chain dependencies, and vendor concentration.
Geopolitical Risk
Risks from political instability, sanctions, trade policy shifts, and cross-border regulatory divergence.
Conduct Risk
Risks from behaviour that results in unfair outcomes for customers, market manipulation, or regulatory censure.
ESG & Climate
Physical and transition risks arising from climate change, nature-related financial risks, and sustainability obligations.
Credit Risk
Risks from counterparty default, concentration, and deterioration in asset quality across lending portfolios.
Market Risk
Risks from adverse movements in interest rates, foreign exchange, equity prices, and commodity prices.
Liquidity Risk
Risks from an inability to fund obligations as they fall due, including intraday and stress liquidity management.
Regulatory Risk
Risks from evolving regulatory requirements, enforcement actions, and supervisory expectations.
Strategic Risk
Risks from poor strategic decisions, misaligned business models, or failure to respond to structural industry change.
Cyber Risk
Risks from cyber attacks, data breaches, ransomware, and vulnerabilities in digital infrastructure.
Source Types
Limitations
Risk Horizon does not provide legal, regulatory, or investment advice. All intelligence outputs should be reviewed by qualified professionals before being relied upon.
AI language models can misinterpret regulatory text, omit important context, or generate plausible-sounding but inaccurate summaries. Critical decisions should be validated against the primary source document.
Coverage is limited to publicly available materials. Firm-specific supervisory correspondence, confidential regulatory discussions, and non-public enforcement proceedings are not included.
Materiality and confidence scores are AI-generated estimates, not actuarial or quantitative assessments. They reflect relative significance within the platform, not absolute risk measurement.
The platform does not constitute a risk management system and should not replace an institution's own risk frameworks, internal governance, or regulatory engagement processes.