| # Configuration file for the agentic AI trading system | |
| data_source: | |
| type: 'yahoo' | |
| path: 'data/market_data.csv' | |
| trading: | |
| symbol: 'AAPL' | |
| timeframe: '1d' | |
| capital: 100000 | |
| risk: | |
| max_position: 100 | |
| max_drawdown: 0.05 | |
| execution: | |
| broker_api: 'paper' # Options: 'paper', 'alpaca_paper', 'alpaca_live' | |
| order_size: 10 | |
| delay_ms: 100 | |
| success_rate: 0.95 | |
| # Alpaca configuration | |
| alpaca: | |
| api_key: '' # Set via environment variable ALPACA_API_KEY | |
| secret_key: '' # Set via environment variable ALPACA_SECRET_KEY | |
| paper_trading: true # Use paper trading by default | |
| base_url: 'https://paper-api.alpaca.markets' # Paper trading URL | |
| live_url: 'https://api.alpaca.markets' # Live trading URL | |
| data_url: 'https://data.alpaca.markets' # Market data URL | |
| websocket_url: 'wss://stream.data.alpaca.markets/v2/iex' # WebSocket URL | |
| account_type: 'paper' # 'paper' or 'live' | |
| # Yahoo Finance (default ingest). Unofficial API, typically delayed; 1m lookback is ~7 days. | |
| # Unofficial API, typically delayed; 1m lookback is ~7 days. | |
| yahoo: | |
| poll_interval_seconds: 60 | |
| # Keep true. With auto_adjust off, Yahoo returns raw Close and every stock | |
| # split reads as a crash (NVDA's June 2024 10:1 becomes a -90% bar). | |
| auto_adjust: true | |
| # Yahoo returns the still-forming period as an ordinary row. Emitting it would | |
| # trade on a close that has not happened yet. | |
| emit_incomplete_bars: false | |
| max_backoff_seconds: 900 | |
| start_date: '2024-01-01' | |
| end_date: '2026-12-31' | |
| # Synthetic data generation settings | |
| synthetic_data: | |
| base_price: 150.0 | |
| volatility: 0.02 | |
| trend: 0.001 | |
| noise_level: 0.005 | |
| generate_data: true | |
| data_path: 'data/synthetic_market_data.csv' | |
| # Logging configuration | |
| logging: | |
| log_level: 'INFO' | |
| log_dir: 'logs' | |
| enable_console: true | |
| enable_file: true | |
| max_file_size_mb: 10 | |
| backup_count: 5 | |
| # FinRL configuration | |
| finrl: | |
| algorithm: 'PPO' # PPO, A2C, DDPG, TD3 | |
| learning_rate: 0.0003 | |
| batch_size: 64 | |
| buffer_size: 1000000 | |
| learning_starts: 100 | |
| gamma: 0.99 | |
| tau: 0.005 | |
| train_freq: 1 | |
| gradient_steps: 1 | |
| target_update_interval: 1 | |
| exploration_fraction: 0.1 | |
| exploration_initial_eps: 1.0 | |
| exploration_final_eps: 0.05 | |
| max_grad_norm: 10.0 | |
| verbose: 1 | |
| tensorboard_log: 'logs/finrl_tensorboard' | |
| training: | |
| total_timesteps: 100000 | |
| eval_freq: 10000 | |
| save_best_model: true | |
| model_save_path: 'models/finrl_best/' | |
| inference: | |
| use_trained_model: false | |
| model_path: 'models/finrl_best/best_model' | |