Inspiration
The challenge required building an automated trading algorithm for 8 different markets based on real-time Munich data. We were inspired to create a data-driven trading system that combines real-time data scraping, fair value calculation, and automated order execution
What it does
Our system monitors real-time metrics (Isar river water levels/flows, Munich weather, airport flight data) via a FastAPI service, calculates fair values for each market product, and automatically executes buy/sell orders when prices deviate from calculated fair values. The bot includes dynamic pricing logic, one-shot order scripts, and order management tools.
How we built it
Built with Python using FastAPI for metrics calculation, BeautifulSoup/Playwright for data scraping, and a custom trading bot framework that connects to the exchange via SSE streams. The system includes parsers for river data, weather forecasts, and flight information, which feed into pricing models that calculate fair values for options and ETF products.
Challenges we ran into
Managing large virtual environment files in git, parsing German-formatted river data tables, handling real-time data synchronization between multiple data sources, and implementing proper rate limiting to respect exchange API constraints. We also had to handle edge cases in formulas which depend on min/max water levels and flow rates over time windows.
Accomplishments that we're proud of
Successfully integrating multiple real-time data sources into a unified pricing API, implementing automated trading logic that responds to price drops, and creating a modular architecture with separate scripts for different trading strategies. We also built robust error handling and logging for production trading.
What we learned
how to structure real-time data pipelines, complexity of pricing derivative products based on multiple underlying data streams, and the importance of teamwork git hygene. We also learned about market-making strategies and risk management in algorithmic trading.
What's next for IMucCity
Enhancing the pricing models with machine learning predictions, implementing more sophisticated risk management, and adding portfolio-level optimization across all 8 markets. We'd also like to add backtesting capabilities and real-time PnL tracking.
