Competition Tracks

“Long Green Cup” National College AI Stock Market Risk Prediction Challenge

Track 1: Multimodal Data-Driven Market Risk Prediction (Core Algorithm Track)

• Background: Traditional stock market risk prediction heavily relies on quantitative price and volume data. However, the current breakthrough in AI lies in its ability to process unstructured data.
• Task: Participating teams are required to utilize Large Language Models (LLMs) to process unstructured data, such as financial news, earnings reports, and social media sentiment. By integrating this with traditional time-series price and volume data, teams must construct a multimodal risk prediction model.
• Evaluation Metrics: The model’s early warning accuracy for extreme risk events (such as flash crashes or consecutive limit-downs), F1-Score, and model inference latency.
• Target Majors: Computer Science and Technology, Artificial Intelligence, Data Science.

Track 2: Dynamic Portfolio and Tail Risk Hedging (Financial Practice Track)

• Background: The ultimate goal of predicting risk is to manage it. This track evaluates the practical implementation capabilities of AI in real-world financial scenarios.
• Task: Given an initial pool of funds containing multiple stocks, participating teams need to utilize AI algorithms for asset allocation and dynamically adjust positions amid market fluctuations to hedge against systemic risks or individual stock “black swan” events.
• Evaluation Metrics: Risk-adjusted returns (such as Sharpe Ratio, Sortino Ratio), maximum drawdown control rate, and performance resilience under extreme market conditions.
• Target Majors: Finance, Financial Engineering, Statistics, Economics.