TradingTeamAI — Autonomous Trading Council
This Ospina case study documents how Carlos Rico-Ospina approached a specific risk, infrastructure, revenue, or research problem and what was built to address it.
Built an AI system that doesn't chat—it debates, checks its math, and enforces risk compliance before every trade.
The Problem
Standard AI chatbots are dangerous for trading. They hallucinate numbers, lack memory, and can't enforce risk limits. Conducting institutional-grade research manually takes hours. Doing it with vanilla LLMs leads to math errors and dangerous position sizing.
The Insight
Successful trading isn't a solo activity—it's a workflow of specialized roles (Macro Economist, Technical Analyst, Risk Manager) with built-in tension. A single LLM prompt can't simulate this. The solution wasn't a chatbot—it was an Autonomous Agent Council where the AI must 'pitch' trades to a strict, code-based risk engine.
What I Built
- Engineered 5 specialized AI agents using CrewAI: Macro Analyst, Technical Analyst, Risk Specialist, Lead Strategist, Documentation Specialist
- Built deterministic Python RiskCalculator tool—the Risk Specialist cannot use LLM math
- Implemented Vector Search via Convex to recall past research notes, solving the 'goldfish memory' problem
- Created structured JSON output with validated logic, checked math, and cited sources
Outcomes
- Reduced institutional-grade trade thesis generation from ~3 hours to minutes
- Zero math hallucinations—risk calculations enforced through deterministic tooling
- Full audit trail with documented reasoning for each decision
- Live architecture demonstrating advanced agentic workflows
Why It Matters
Most AI wrappers 'chat.' This system enforces compliance with hard-coded safety rails.
Proves you can build AI systems that operate within financial safety bounds.
Related Projects
View All
AI / MLQuantitative Risk Engine
2024Built an ML-driven risk platform that processes 89 million trades and identifies dangerous behavior before blowouts happen.
AI / MLWeather Market Probabilistic Forecasting
2025Built two probabilistic models that identify conditions where forecast skill beats market consensus—turning calibrated edge into directional trading signals.
Have a Similar Challenge?
Let's discuss how I can help you achieve similar results.
Start a Conversation