CoTrade

Inspiration

For years, Members of Congress have consistently outperformed the market — not because they’re financial geniuses, but because they have privileged access to information and influence over policy. The public gets access to these filings, but usually too late, and most tools just dump raw data without context. While existing tools surface raw disclosure feeds, they fail to deliver actionable intelligence or connect those disclosures to real-time trading decisions. We wanted to close that loop by building an end-to-end system that contextualizes these trades, manages risk, and can ultimately execute autonomously.

We built CoTrade to flip the script: to take the same information Congress uses to grow wealth and make it accessible, contextualized, and actionable for everyone.

What it does

CoTrade AI ingests real-time financial disclosures, enriches them with contextual signals (such as legislative influence and sector exposure), and translates them into structured, risk-aware trading insights. The system is designed to:

• Detect and scrape disclosures deterministically.
• Enrich and contextualize disclosures with LLM-based reasoning.
• Store and serve structured insights via robust AWS data pipelines.
• Give users most recent trades based on their preferences.

How we built it

We architected CoTrade entirely on AWS-managed services to ensure scalability and reliability.

• Data ingestion: Deterministic scraping and AWS pipelines for real-time disclosures.
• Storage & retrieval: Managed data stores with retrievers to connect signals.
• Contextualization: LLMs running on AWS services to enrich disclosures with legislative and market context.
• System orchestration: Modular architecture enabling future integrations with broker APIs.

Challenges we ran into

• Designing reliable scrapers for inconsistent and evolving disclosure formats.
• Balancing determinism with the probabilistic nature of LLMs.
• Integrating multiple AWS-managed services smoothly without introducing overhead.

Accomplishments that we're proud of

• Building an end-to-end pipeline from disclosure detection to contextualized trading insight.
• Creating a framework that exposes and leverages the same trading edge Congress enjoys.
• Showing how AI can turn opaque disclosures into transparent market intelligence.

What we learned

• How to combine deterministic scraping with LLM-driven contextualization effectively.
• Best practices for orchestrating data pipelines and retrieval systems on AWS.
• The challenges of bridging regulatory data sources with real-world trading systems.

What's next for CoTrade

• Implementing an API that directly connects CoTrade to trading platforms, enabling autonomous, risk-managed execution.
• Expanding coverage to other high-impact market actors beyond Congress.
• Advanced contextualization with fine-tuned financial and political AI models.

Built With

• amazon-web-services
• bedrock