EU AI Act (Regulation 2024/1689), Article 52
Per Annex III, high-risk AI includes systems for creditworthiness assessment. Kareg does not assess individual creditworthiness — it provides macro-level market regime classification that applies uniformly to all market participants.
Every component of the pipeline is documented and auditable.
| Component | Method | Transparency |
|---|---|---|
| Feature computation | Rolling z-scores (multi-year window) | Fully documented |
| Classification | Threshold-based state machine | Thresholds published |
| Severity assessment | Weighted composite score | Weights published |
| Decision (policy) | Rule-based lookup table | Reproducible via audit |
Controls that keep humans in the loop at every step.
Each daily signal reviewed before publication
Manual override via dispute mechanism
Full provenance: input data → features → classification → decision
Pipeline can be halted at any stage
No personal data. No profiling. GDPR not applicable.
Not a benchmark under BMR Article 3(1)
| US macro sources | Redundant government data APIs |
| EU macro sources | Alternative official statistical providers |
| Market data providers | Direct exchange feeds as fallback |
Benchmark Regulation Notice: Kareg regime signals are advisory risk overlays and do not constitute a benchmark within the meaning of Regulation (EU) 2016/1011. The signals are derived from publicly available macroeconomic data and should not be used as the sole basis for investment decisions or financial instrument valuation.
Retention policy compliant with Article 12 (ICT security and data integrity).