How luckyone7777/chatgpt-micro-cap-experiment works
Project type and purpose
• This is a Python-based data-pipeline/automation tool rather than a classic client-server “full-stack” app. It’s an experiment framework where a decision engine (ChatGPT) consumes market data each trading day, applies rules, emits trade instructions, logs them to CSV, and generates performance visualizations.
File structure
Top-level directories and files:
• Experiment Details – research artifacts, prompts, Q&A, disclaimers.
• Scripts and CSV Files – core scripts (Generate_Graph.py, Trading_Script.py) plus generated CSV logs (chatgpt_portfolio_update.csv, chatgpt_trade_log.csv).
• Start Your Own – a “template” copy of the core scripts and CSV files for bootstrapping a new experiment.
• Weekly Deep Research (MD/PDF) – weekly summary documents in Markdown and PDF.
• trading_script.py (root) – duplicate/entry-point for live trading logic.
• requirements.txt – Python dependencies (pandas, yfinance, matplotlib, OpenAI/ChatGPT SDK).
Architectural choices:
• Modular scripts separate data ingestion + decision logic (trading_script.py) from visualization (Generate_Graph.py).
• CSV files as simple persistent storage.
• “Start Your Own” directory as a scaffold for new experiments.
README for additional insights
• Tech stack: Python, pandas + yfinance (market data), Matplotlib (visuals), ChatGPT (trade decisions).
• Workflow: daily data feed → ChatGPT decision engine → strict stop-loss + trade generator → CSV log updates → weekly performance graphs.