title	emoji	colorFrom	colorTo	sdk	app_port	pinned
AutoTrader	📈	green	blue	docker	7860	false

AutoTrader - Autonomous Paper Trading Bot

AI-powered paper trading bot that scans high-volume stocks, analyzes them with RSI and news sentiment via LLM reasoning, and executes paper trades on Alpaca.

→ STRUCTURE.md — Project layout and navigation

Features

• Stock Scanner: Identifies top 50 most active stocks by volume
• Technical Analysis: RSI (14-period) using Wilder's smoothing
• News Sentiment: Fetches recent headlines via Alpaca News API
• LLM Reasoning: Uses Llama 3.3 70B via HF Inference API for analysis
• Paper Trading: Executes trades on Alpaca paper trading
• Risk Management: 5% max position size, confidence threshold
• Dashboard: Real-time Gradio UI with account summary, positions, and trade history
• Scheduling: Automatic scans every 15 minutes during market hours

Required Secrets

Secret	Description
ALPACA_API_KEY	Alpaca paper trading API key
ALPACA_SECRET_KEY	Alpaca paper trading secret key
HF_TOKEN	Hugging Face token (for Inference API)

Local Development

cp .env.example .env
# Fill in your API keys in .env
pip install -r requirements.txt
python app.py

Dashboard will be available at http://localhost:7860.

Architecture (refactor)

• Single scan entrypoint: workspace/scan_autotrader.py — used by cron and HEARTBEAT; uses shared lib (in-process Alpaca client, retries, logging).
• Shared lib (workspace/lib/): config (watchlist, env validation), alpaca_client (get_account, get_positions, get_bars, get_snapshot, buy, sell with retries), rsi, decisions (log, retention, outcomes, daily review).
• Watchlist: workspace/config/watchlist.json — single source of ticker groups.
• Self-improvement: Each scan appends to logs/outcomes.jsonl and logs/daily_review.jsonl; logs/decisions.jsonl is rotated (90-day retention). See workspace/SELF_IMPROVEMENT.md.
• Health: GET /api/health checks Alpaca connectivity.
• Discord: Set DISCORD_BOT_TOKEN in .env; do not store the token in openclaw-config/openclaw.json. See DISCORD.md.

Test before market open

1. Health check: Start the stack, then curl http://localhost:5050/api/health (or open in browser). Should return {"alpaca":"ok", "equity": ...}.
2. Scan dry run: From repo root, docker compose exec openclaw-gateway python /home/node/.openclaw/workspace/scan_autotrader.py (uses container env; will hit Alpaca).
3. Discord: Ensure Message Content Intent is ON and DISCORD_BOT_TOKEN is in .env; restart gateway after any config change.