EDT

The EDT dataset is designed for corporate event detection and text-based stock prediction (trading strategy) benchmark.

1. Corporate Event Detection It includes 9721​ news articles with token-level event labels. Including 11 event types: Acquisitions, Clinical Trials, Guidance Changes, New Contracts, Stock Repurchases, Stock Split, Reverse Stock Split/ADS Ratio Change, Regular Dividend, Special Dividend, Dividend Cut, Dividend Increase

2. Text-Based Stock Prediction Benchmark It includes 303893​ first-hand news articles from high-quality sources. Each news article is assigned a minute-level timestamp and comprehensive stock price labels.

Please see this Github Link and paper for more details.