The Two-Stage Leaderboard (Robustness Check)

We are simulating a real hedge fund environment.

  • Public Leaderboard (30% of Data): Visible immediately. Use this to tune your model.

  • Private Leaderboard (70% of Data): Hidden until the competition ends.

  • Warning: If you overfit your model to the Public data, you will fail the Private test. We reward Generalization, not memorization.

2. Eligibility

  • Open to all Students, and Students Only

  • Teams of 1 to 4 members.

  • University students may participate as "Mentors" but are ineligible for most prizes except for participant prizes.

3. Code Requirements

  • No Manual Trading: All signals must be generated by an automated script.

  • Open Source: Winners must open-source their kernels for verification to claim prizes.

  • No Look-Ahead Bias: Using future data to predict past events (data leakage) results in immediate disqualification.