# AI Trading Journey

> 90-day experiment building an autonomous AI trading system. Real failures, real fixes, real lessons learned.

## Quick Summary
- **Phase**: R&D Day 50/90
- **Mode**: Paper Trading (no real money)
- **Goal**: Fibonacci compounding ($1/day scaling)
- **Stack**: Python, Alpaca API, Claude AI, PyTorch

## Daily Reports
- [Latest Report](https://igorganapolsky.github.io/trading/reports/): Most recent daily P/L and trade analysis
- [Performance Log](https://github.com/IgorGanapolsky/trading/blob/main/data/performance_log.json): Raw JSON performance data

## Lessons Learned
- [All Lessons](https://igorganapolsky.github.io/trading/lessons/): 60+ documented failures and fixes
- [Critical Lessons](https://github.com/IgorGanapolsky/trading/tree/main/rag_knowledge/lessons_learned): Markdown source files

## Trading Strategies
- **5-Tier Strategy**: MACD + RSI + RL + Sentiment + Risk Management
- **Asset Classes**: US Equities and Options (NO crypto)
- **Market Hours**: Mon-Fri 9:30 AM - 4:00 PM ET only

## Key Metrics
- Win Rate Target: >60%
- Sharpe Ratio Target: >1.5
- Max Drawdown Limit: -5%
- Position Size: Kelly Criterion with 25% safety margin

## Risk Management
- Circuit breakers for rapid loss prevention
- Position limits per symbol and strategy
- Volatility-adjusted sizing
- Pre-trade risk assessment

## Repository
- [GitHub](https://github.com/IgorGanapolsky/trading): Full source code
- [Dev.to Series](https://dev.to/igorganapolsky): Blog posts and updates

## Contact
- Author: Igor Ganapolsky
- CTO: Claude AI (autonomous execution)
- Twitter: @igorganapolsky

---
*Educational purposes only. Not financial advice.*