Paper: arXiv 2411.08726

Abstract

This article applies natural language processing (NLP) to extract and quantify textual information to predict stock performance. Using an extensive dataset of Chinese analyst reports and employing a customized BERT deep learning model for Chinese text, this study categorizes the sentiment of the reports as positive, neutral, or negative. The findings underscore the predictive capacity of this sentiment indicator for stock volatility, excess returns, and trading volume. Specifically, analyst reports with strong positive sentiment will increase excess return and intraday volatility, and vice versa, reports with strong negative sentiment also increase volatility and trading volume, but decrease future excess return. The magnitude of this effect is greater for positive sentiment reports than for negative sentiment reports. This article contributes to the empirical literature on sentiment analysis and the response of the stock market to news in the Chinese stock market.

Complexity vs Empirical Score

  • Math Complexity: 4.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper employs an advanced deep learning model (BERT) which carries moderate mathematical complexity, but the core contribution is empirical—using a large proprietary dataset of Chinese analyst reports to perform sentiment analysis and regression on real market data (excess returns, volatility, volume).

Research Flowchart

  flowchart TD
  A["Research Goal<br>Predict Stock Performance in China"] --> B["Data<br>Chinese Analyst Reports"]
  B --> C["Methodology<br>Customized BERT Model for Chinese NLP"]
  C --> D["Process<br>Sentiment Classification<br>Positive / Neutral / Negative"]
  D --> E["Outcomes<br>Sentiment as Predictive Indicator"]
  E --> F["Key Finding 1: Positive Sentiment<br>↑ Excess Return, ↑ Volatility"]
  E --> G["Key Finding 2: Negative Sentiment<br>↓ Excess Return, ↑ Volatility & Volume"]

Cited References (45)

  1. The reverse volatility asymmetry in Chinese financial market (2014)
  2. Asymmetric volatility and risk in equity markets (2000)
  3. Which sentiment index is more informative to forecast stock market volatility? Evidence from China (2020)
  4. The information content of analyst recommendation revisions—Evidence from the Chinese stock market (2014)
  5. The impact of sentiment and attention measures on stock market volatility (2020)
  6. Investor sentiment and stock volatility: New evidence (2022)
  7. Comprehensive review of text-mining applications in finance (2020)
  8. Fad or future? Automated analysis of financial text and its implications for corporate reporting (2019)
and 37 more
  1. Evidence on the Information Content of Text in Analyst Reports (2014) — link
  2. The Predictive Content of Aggregate Analyst Recommendations (2009)
  3. Characteristics of a firm’s information environment and the information asymmetry between insiders and outsiders (2004)
  4. Who reacts to what information in securities analyst reports? Direct evidence from the investor trade imbalance (2021)
  5. The information content of analysts’ textual reports and stock returns: Evidence from China (2022)
  6. Information uncertainty, investor sentiment, and analyst reports (2021)
  7. Giving content to investor sentiment: The role of media in the stock market (2007)
  8. Twitter mood predicts the stock market (2011)
  9. The impact of Twitter-based sentiment on US sectoral returns (2023)
  10. Information content of equity analyst reports (2005)
  11. The predictive content of aggregate analyst recommendations (2009)
  12. Evidence on the information content of text in analyst reports (2014)
  13. Investor inattention and the underreaction to stock recommendations (2010)
  14. Stock-uniBERT: a news-based cost-sensitive ensemble BERT model for stock trading (2020)
  15. Sentiment analysis of Chinese stock reviews based on BERT model (2021)
  16. Persistence of volatility in futures markets (2006)
  17. Testing range estimators of historical volatility (2006)
  18. On the estimation of security price volatilities from historical data (1980)
  19. Distillation of news flow into analysis of stock reactions (2016)
  20. Analyzing analyst research: A review of past coverage and recommendations for future research (2018)
  21. National culture and analysts’ forecasting (2023)
  22. The information content of earnings announcements: new insights from intertemporal and cross-sectional behavior (2018)
  23. Market segmentation and differential reactions of local and foreign investors to analyst recommendations (2017)
  24. Cultural proximity and the processing of financial information (2017)
  25. Pre-Training With Whole Word Masking for Chinese BERT (2021)
  26. Textual analysis for China’s financial markets: a review and discussion (2020)
  27. Analyst information discovery and interpretation roles: A topic modeling approach (2018)
  28. Finbert: A pretrained language model for financial communications (2020) — link
  29. Bert: Pre-training of deep bidirectional transformers for language understanding (2018) — link
  30. Optimistic bias of analysts’ earnings forecasts: Does investor sentiment matter in China? (2018)
  31. Limited investor attention and biased reactions to information: Evidence from the COVID-19 pandemic (2023)
  32. Investor sentiment, style investing, and momentum (2023)
  33. Media abnormal tone, earnings announcements, and the stock market (2022)
  34. Investor sentiment and the cross-section of stock returns (2006)
  35. The short of it: Investor sentiment and anomalies (2012)
  36. Investor sentiment aligned: A powerful predictor of stock returns (2015)
  37. What reflects investor sentiment? Empirical evidence from China (2021)

0 of 45 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.