Premise
A serious market workflow needs more than a trading bot. Model the organization as separate operating layers, build the administrative foundation first, and prevent the agent generating an idea from approving its own work.
Setup
- Map required legal, billing, recordkeeping, and compliance work before market research.
- Define eight narrow roles across operations, research, signals, risk, simulated execution, and growth.
- Pair every maker with an independent checker.
- Ground research methods in citable papers and verifiable data.
- Keep execution connected only to simulation.
Boundaries
- No live orders, outside capital, investor solicitation, or performance marketing.
- Entity formation and legal conclusions require qualified humans.
- Unsourced data and backtests cannot enter a recommendation.
Begin with the operational layer. Produce a checklist for entity, accounting, billing, recordkeeping, and compliance needs, clearly separating automatable administration from work requiring legal or financial professionals. Stop for approval before designing research.
Next, propose reproducible research methods with primary citations and real data requirements. Ensure every generated signal goes to a different checker that examines evidence and risk before it reaches me. No role may approve its own output.
Execution is paper-only. Do not form an entity, take payments, accept or solicit capital, claim performance, connect a live brokerage, or place orders. Label market material as research rather than financial advice, and report unavailable evidence instead of fabricating it.
Prompt
Help me design an eight-role research and operations desk across business administration, research, signal generation, independent review, risk, simulated execution, and growth.