Operating Model¶
Note
This document describes the target operating model. Not all steps are fully implemented in the current codebase. The system is an experimental scaffold under active development.
12-Step Operating Loop¶
graph LR
I[1. Ingest] --> N[2. Normalize]
N --> C[3. Classify Regime]
C --> G[4. Generate Candidates]
G --> F[5. Forecast]
F --> V[6. Cross-Validate]
V --> R[7. Rank]
R --> P[8. Construct Portfolio]
P --> RK[9. Risk Validation]
RK --> E[10. Execute]
E --> M[11. Monitor]
M --> REV[12. Review & Evolve]
REV -.-> I
- Ingest — Market, macro, fundamental, technical, and alternative data
- Normalize — Enrich data into canonical entities
- Classify — Market regime, liquidity state, and volatility context
- Generate — Candidate opportunities across strategies and assets
- Forecast — Distributions, scenario ranges, and confidence
- Cross-validate — Candidate theses across specialist agents
- Rank — Opportunities by expected return, confidence, liquidity, and regime fit
- Construct — Portfolio under portfolio-level constraints
- Risk validate — Pre-trade risk checks and veto
- Execute — Cost-aware routing and scheduling
- Monitor — Fills, exposure, P&L, health, and incidents in real time
- Review — Attribution, drift detection, challenger tests, and evolution under governance
Governance Model¶
Humans define the operating boundaries:
- Mandates — Allowed assets, strategies, allocation caps
- Risk budgets — Drawdown limits, leverage ceilings, exposure caps
- Compliance constraints — Instrument restrictions, trading hours
- Override conditions — Kill switch, manual pause, forced deleverage
AIS operates within these boundaries and escalates incidents or policy breaches.
Current Implementation Status¶
| Step | Status | Module |
|---|---|---|
| Ingest | Implemented | data/, exchange/ |
| Normalize | Implemented | data/providers/ |
| Classify | Partial | agents/market_intelligence/ |
| Generate | Implemented | agents/strategy/ |
| Forecast | Partial | quant/ |
| Cross-validate | Via arbitration | orchestration/ |
| Rank | Implemented | orchestration/arbitration.py |
| Construct | Implemented | portfolio/ |
| Risk validate | Implemented | risk/ |
| Execute | Implemented | execution/ |
| Monitor | Implemented | monitoring/ |
| Review | Implemented | review/ |