Reproduce “Towards Earnings Call and Stock Price Movement” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhiqiang Ma, Grace Bang, Chong Wang, Xiaomo Liu
Researchers from S&P Global developed a deep learning model that predicts the directional movement of stock prices (up or down) following earnings calls, achieving 52.45% accuracy and a Matthews Correlation Coefficient of 0.0445. The model utilizes the Q&A sections of earnings call transcripts and company industry classifications, demonstrating that unscripted management commentary provides predictive power for post-announcement stock performance.
Key insights
- The 'Answer' sections of earnings calls, representing dynamic Q&A, possess more predictive power for future stock price direction than the scripted 'Presentation' sections.
- The model's predictive performance varies across industries; for example, Information Technology stocks are more predictable from earnings calls than Energy stocks.
- Combining advanced natural language processing with industry-specific embeddings offers a superior approach to predicting stock price direction compared to traditional baseline methods.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.