题　目：Deepseeking Investment Value through Analyst Reports

嘉　宾：李伟凯 副教授 香港城市大学

主持人：万相伟 副教授 上海交通大学安泰经济与管理学院

时　间：2026年5月20日（周三）13:30-15:00

地　点：上海交通大学 徐汇校区安泰浩然楼306

内容简介:

We investigate the incremental informational content of qualitative opinions embedded in a comprehensive sample of Chinese analyst reports, as extracted by the DeepSeek large language model. We construct firm-level sentiment measures using DeepSeek’s advanced Chinese-language processing capabilities and find that analyst sentiment significantly predicts cross-sectional stock returns and future firmfundamentals, even after controlling for analyst quantitative outputs, including recommendation changes and earnings forecast revisions. The predictive power of sentiment persists with alternative measurement methods, at the industry level, and in mutual fund performance analyses. These results provide strong evidence that large language models can systematically extract value-relevant information from analyst narratives, thereby advancing both the literature on financial textual analysis and the information intermediary role of analysts.

演讲人简介：

香港科技大学金融学博士，特许金融分析师(CFA)，香港城市大学金融学（终生）副教授。研究领域涉及实证资产定价、行为金融以及可持续金融。现担任Pacific Basin Finance Journal 和Asia-Pacific Journal of Financial Studies 副主编。曾在香港金融管理局香港货币及金融研究中心 (HKIMR)担任访问研究员。在Review of Financial Studies，Journal of Financial Economics，Journal of Financial and Quantitative Analysis，Journal of Econometrics, Review of Accounting Studies等国际学术期刊上发表多篇学术论文。研究曾在中国金融国际年会(CICF)，国际金融管理协会年会(FMA Asia & FMA Europe)，亚洲金融协会年会 (Asian FA)，芝加哥数量投资协会 (CQAsia)以及国际可持续金融研究联盟（GRASFI）等国际学术会议上获奖。曾多次在中投公司、城堡投资、美银美林等国内外金融机构宣讲论文，研究成果被《华尔街日报》，《金融时报》和《清华金融评论》等报道转载。

欢迎广大师生参加!