This project is my attempt to build a simple automated workflow for quantitative research. It’s not meant to replace professional tools — the goal is to learn how quant research actually works by building all steps myself.

The system has four small agents:

1. Idea Agent

   • Creates simple trading ideas (mean reversion, momentum, etc.)

2. Data Agent

   • Downloads the price data needed for testing.

3. Backtest Agent

   • Tests each idea over the historical data.

4. Analyzer Agent

   • Checks which idea performed best and summarizes the results.

Then there’s an Orchestrator which just runs them in order.

I wrote this project to practice:

• time-series data handling
• backtesting logic
• basic metrics (Sharpe, drawdown)
• modular Python design
• building tools instead of copying them