摘要
We develop market-level analyst opinion indices using detailed sentiment analysis of summary texts from analyst reports, employing large language models (LLMs) and various aggregation techniques such as partial least squares (PLS), scaled principal component analysis (sPCA), and principal component analysis (PCA). Our results indicate that these aggregated opinion indices can significantly and positively predict monthly market returns, both in sample and out of sample. The positive predictive power of our indices appears to stem from an underreaction mechanism, whereby information in analyst reports is not immediately or fully incorporated into prices. This research enhances the understanding of stock return predictability and underscores the potential of AI in improving financial analysis.
| 源语言 | 英语 |
|---|---|
| 期刊论文编号 | 107688 |
| 期刊 | Journal of Banking and Finance |
| 卷 | 187 |
| DOI | |
| 出版状态 | 已出版 - 6月 2026 |
| 已对外发布 | 是 |
学术指纹
探究 'Can analyst textual opinions predict stock returns?' 的科研主题。它们共同构成独一无二的学术指纹。引用此
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver