Reinforcement Learning
RL Portfolio Optimizer
Deep reinforcement learning for portfolio allocation. PPO agent achieving 84.6% CAGR vs 47.1% baseline.
About the Project
This project uses deep reinforcement learning to optimize multi-asset portfolio allocation. A PPO agent learns to dynamically allocate capital across tech stocks (AAPL, TSLA), cryptocurrencies (BTC, ETH), and ETFs (SPY) based on 50 technical features per asset. The system significantly outperforms traditional equal-weight strategies through learned market timing and risk management.
Key Features
- PPO (Proximal Policy Optimization) agent for portfolio allocation
- 50 technical features per asset including SMA, EMA, RSI, volatility
- Trained on 2018-2024 historical data across crypto and stocks
- 84.6% CAGR vs 47.1% equal-weight baseline (+80% outperformance)
- 1.33 Sharpe ratio with TensorBoard monitoring
Impact
Achieved 84.6% CAGR with 1.33 Sharpe ratio, outperforming the equal-weight baseline by 80%. Demonstrates practical application of RL in quantitative finance.
Tech Stack
PyTorchStable-Baselines3GymnasiumyfinanceTensorBoard
Metrics
1.33 Sharpe ratio
50 features/asset
Multi-asset: AAPL, BTC, ETH, TSLA, SPY
Interested in this project?
Let's discuss how I can build something similar for you.