Run trading bots without running servers. Pick a strategy, set your limits, and watch a bot trade a simulated wallet on live Kraken or Binance.US market data. A portfolio project, paper trading only: no exchange keys, no real orders, no real money.
Django and Django REST Framework with JWT auth, Celery workers and beat on Redis, PostgreSQL, and a candle-cache proxy in front of the exchanges. Every bot is a freqtrade container on an isolated Docker network with CPU and memory caps, driven through its REST API and reconciled every minute. Strategy parameters are stored as JSON schema and rendered as forms; 30-day backtests give each strategy its profit, win rate and drawdown. Telegram alerts, subscription tiers, Traefik with Let's Encrypt on a single VPS sized for a demo, and three test suites (Django, candle cache, Playwright end to end) in CI. The write-up says what this design is good for and what would change at scale.
Built end to end with Claude Code as a pair programmer: scoped tasks, every diff reviewed, tests as the gate on money math and order simulation.
- Python
- Django
- DRF
- Celery
- PostgreSQL
- Redis
- Docker
- Traefik
- freqtrade
- Playwright