Swingbot

Customizable swing trading bot template. Scans a watchlist on a schedule, evaluates a pluggable strategy, posts signals to Telegram + Discord, and optionally executes paper trades on Alpaca. Built to be cloned, customized, and deployed in under an hour.

What it does

For each symbol in config.yaml, swingbot fetches daily bars, runs the strategy module's evaluate() function, and reports signals. In scan mode (the default) it just notifies. In enter mode it sizes the position based on configured risk and submits a market order. In manage mode it checks open positions for stop or target hits and closes them.

# Example config.yaml
watchlist: [SPY, QQQ, AAPL, MSFT, NVDA]
risk_pct_per_trade: 1.0
strategy:
  sma_fast: 20
  sma_slow: 50
  atr_stop_mult: 1.5
  risk_reward: 2.0

Architecture

┌──────────────────────┐
│ GitHub Actions cron  │   open + pre-close
└──────────┬───────────┘
           │
           ▼
┌──────────────────────────────────┐
│ swingbot.py (orchestration)      │
│  --mode scan  | enter | manage   │
└──┬─────────┬────────────┬────────┘
   │         │            │
   ▼         ▼            ▼
config    strategy.py  broker.py
.yaml     (signals)    (Alpaca)
                          │
                          ▼
                       Alpaca
                       (paper)
            ┌─────────────┘
            ▼
       notifications.py
            │
   ┌────────┴────────┐
   ▼                 ▼
Telegram         Discord

Modular by design — strategy.py is the only file you need to touch to plug in a different system (RSI divergence, Bollinger bands, ICT setups, machine-learning signals — anything that returns a Signal or None).

Modes

Mode	Behavior
scan	Read-only — fetches bars, evaluates strategy, notifies
enter	Scan + submits Alpaca paper market orders for new signals
manage	Checks open positions for stop/target hits, closes them

The included workflow runs scan by default. To go live, change --mode scan to --mode enter in .github/workflows/cycle.yml and add a separate manage cycle if desired.

Safety

• Defaults to paper=True on Alpaca — no live brokerage by default
• Default cycle is scan only — won't place orders without explicit opt-in
• Position sizing risk is capped via risk_pct_per_trade in config
• All actions logged to GitHub Actions output and notified to Telegram

Quick start

1. Fork or clone this repo
2. Edit config.yaml with your watchlist and strategy parameters
3. Set repo secrets:
   • ALPACA_API_KEY and ALPACA_SECRET_KEY from alpaca.markets (free paper account)
   • TELEGRAM_BOT_TOKEN, TELEGRAM_CHAT_ID
   • DISCORD_WEBHOOK_URL (optional)
4. Commit and push
5. Run the workflow manually first to verify — should report scan results to Telegram

Custom builds

Need this extended — different broker (IBKR, Tradier, TradeStation), live brokerage execution, advanced indicators, multi-timeframe analysis, custom risk management, Claude-reviewed signal filtering, multi-account routing? I build production trading systems for retail traders, prop desks, and algorithmic firms.

Built by Stillrun Lab — automation systems built to run themselves.

• 💼 Hire me on Upwork: (link coming soon)
• 🐦 @trade_4l on X
• 📧 (email coming soon)

Disclaimer

This is template code for educational and paper-trading purposes. It does not constitute financial advice. The included strategy is illustrative; backtest and paper-trade extensively before deploying capital.