# alpaca-llm-bot-v1

Autonomous LLM trading system (paper-first) powered by:
- Alpaca trading API
- Ollama local model inference
- Searx web data ingestion
- FastAPI dark dashboard
- APScheduler autonomous loops

## Autonomous behavior
- Curates market/news context every `CURATE_INTERVAL_MINUTES` (default 30)
- Runs trade decision cycle every `TRADE_INTERVAL_HOURS` (default 2)
- LLM decides buy/sell/hold per symbol
- Executes only if confidence >= `MIN_CONFIDENCE`
- Hard risk caps:
  - max `$5` order notional (default)
  - max daily notional cap
  - max open positions cap
  - market-open gate

## Quick start
```bash
cp .env.example .env
# fill ALPACA_API_KEY / ALPACA_API_SECRET
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn app:app --host 0.0.0.0 --port 8089
```

Dashboard:
- `http://<host>:8089/`

Manual triggers:
```bash
curl -X POST http://127.0.0.1:8089/curate-now
curl -X POST http://127.0.0.1:8089/run-now
```

## Production service (systemd)
```bash
sudo cp systemd/alpaca-llm-bot-v1.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now alpaca-llm-bot-v1
sudo systemctl status alpaca-llm-bot-v1
```

## Important
- Keep `PAPER_MODE=true` until you observe stable behavior.
- This is experimental and not financial advice.