MIG Quant
Algo Challenge
The Michigan Investment Group's algorithmic trading challenge. Build a data-driven strategy. Backtest against real market data. Compete against the best minds in university finance.
Process
How It Works
Download Market Data
Access 4 years of historical OHLCV data for 10 fake stock symbols. Data provided in easy-to-use CSV format. Perfect for backtesting your strategy locally.
Code Your Strategy
Implement get_actions(prices) → positions in Python. Use any of the allowed packages and provided algo restraints.
Submit & Compete
Upload your .py file (and model weights for any trained ml model). Backtesting runs automatically. Results appear on the public leaderboard within minutes.
Scoring Structure
Public Leaderboard.
Hidden Finals.
During the submission window, your strategy is evaluated on a public validation dataset. Rankings are visible to everyone — use them to iterate and improve.
After the deadline, every qualifying strategy is re-evaluated on a hidden out-of-sample dataset. Final rankings are determined solely by hidden-set performance. Overfitting is penalized.
Test and develop with your public dataset
Out-of-sample evaluation to generatle our leaderboard.
Final hidden dataset to evaluate the final winners
Presented By
Interested in sponsoring? Contact mig.quant.board@umich.edu