The Challenge
A trading firm operated with "old-school" data analysis tools and wanted to explore how Claude Code or similar agents can help them carry out ongoing research.
Our Approach
We created rich context for a Claude Code agent to serve as an effective trading strategy research assistant.
📊
Market Analysis
Real-time order book parsing
💰
Wallet Tracking
On-chain flow analysis
📈
Price Feeds
Multi-source aggregation
🔗
Trading Infra
Custom MCP integration
🔍
Arbitrage Scan
Cross-platform spreads
⚡
Auto Workflows
Triggered research tasks
Key deliverables:
- Studied the workflow of the researchers and created useful Claude Code skills for prediction market analysis
- Picked the best MCPs for Polymarket/prediction markets integrations
- Integrated several automated workflows, such as wallet analysis-by-address, price feed comparisons, etc
- Created an MCP server to connect Claude Code to the company's own trading infra
- Ran a workshop for the company on how to leverage agentic AI to their advantage
Results
Research Task Speed
Before5 hours
After2 hours
60%faster
- The agent prompted discovery of real trading opportunities
- 60% speed increase in research tasks for the team
- Claude Code works as a viable research assistant
- The agent is infinitely extendable with more skills