Paper: arXiv 2404.03792

Abstract

I explore the relationship between investor emotions expressed on social media and asset prices. The field has seen a proliferation of models aimed at extracting firm-level sentiment from social media data, though the behavior of these models often remains uncertain. Against this backdrop, my study employs EmTract, an open-source emotion model, to test whether the emotional responses identified on social media platforms align with expectations derived from controlled laboratory settings. This step is crucial in validating the reliability of digital platforms in reflecting genuine investor sentiment. My findings reveal that firm-specific investor emotions behave similarly to lab experiments and can forecast daily asset price movements. These impacts are larger when liquidity is lower or short interest is higher. My findings on the persistent influence of sadness on subsequent returns, along with the insignificance of the one-dimensional valence metric, underscores the importance of dissecting emotional states. This approach allows for a deeper and more accurate understanding of the intricate ways in which investor sentiments drive market movements.

Complexity vs Empirical Score

  • Math Complexity: 3.5/10
  • Empirical Rigor: 6.0/10 (capped from 7.0: no out-of-sample test; see penalty flags)
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper uses advanced text analysis and fixed effects econometrics, but lacks the heavy mathematical derivations or complex models typical of high math complexity. It demonstrates strong empirical rigor with a large dataset (88 million messages), backtest-ready methodologies (predictive daily returns), and robustness checks.

Research Flowchart

  flowchart TD
  A["Research Goal:<br/>Test if social media emotions<br/>align with lab-based expectations<br/>and predict asset prices"] --> B["Data Inputs:<br/>Social Media Data<br/>(Firm-Specific Posts)"]
  B --> C["Methodology:<br/>Apply EmTract Model<br/>to extract multi-dimensional<br/>investor emotions"]
  C --> D{"Computational Process:<br/>Test alignment with lab experiments<br/>and forecast daily asset prices"}
  D --> E["Key Outcomes & Findings:<br/>1. Emotions align with lab validation<br/>2. Predict daily asset price movements<br/>3. Impacts amplified by low liquidity/high short interest<br/>4. Sadness drives persistence; simple valence is insignificant"]

Cited References (87)

  1. The role of dissemination in market liquidity: Evidence from firms’ use of Twitter™ (2014)
  2. Animal spirits: How human psychology drives the economy, and why it matters for global capitalism (2010)
  3. General Theory of Employment, Interest, and Money (1936)
  4. Distilling the knowledge in a neural network (2015) — link
  5. Social Media Emotions and IPO Returns (2023) — link
  6. EmTract: Extracting Emotions from Social Media (2023)
  7. Model compression (2006)
  8. An Analysis of Annotated Corpora for Emotion Classification in Text (2018) — link
and 79 more
  1. Bert: Pre-training of deep bidirectional transformers for language understanding (2018) — link
  2. Attention is all you need (2017)
  3. Do firms strategically disseminate? Evidence from corporate use of social media (2018)
  4. DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter (2019) — link
  5. The role of social media in the capital market: Evidence from consumer product recalls (2015)
  6. Racial disparities in debt collection (2019) — link
  7. When the underwriter is the market maker: An examination of trading in the IPO aftermarket (2000)
  8. Hot markets, investor sentiment, and IPO pricing (2006)
  9. Why has IPO underpricing changed over time? (2004)
  10. The new issues puzzle (1995)
  11. Twitter mood predicts the stock market (2011)
  12. Correlating S&P 500 stocks with Twitter data (2012)
  13. Investor attention and the pricing of earnings news (2014)
  14. In search of attention (2011)
  15. Investor information demand: Evidence from Google searches around earnings announcements (2012)
  16. Stock-Price Effects of Internet Buy-Sell Recommendations: The Motley Fool Case (2000)
  17. News or noise? Internet postings and stock prices (2001)
  18. Is all that talk just noise? The information content of internet stock message boards (2004)
  19. Yahoo! for Amazon: Sentiment extraction from small talk on the web (2007)
  20. The value of crowdsourced earnings forecasts (2016)
  21. Wisdom of crowds: The value of stock opinions transmitted through social media (2014)
  22. Can Twitter help predict firm-level earnings and stock returns? (2018)
  23. Good day sunshine: Stock returns and the weather (2003)
  24. Winter blues: A SAD stock market cycle (2003)
  25. Sports sentiment and stock returns (2007)
  26. Widespread worry and the stock market (2010)
  27. Emotional state and market behavior (2018)
  28. Bubbles, crashes, and endogenous expectations in experimental spot asset markets (1988)
  29. Bubbling with excitement: an experiment (2016)
  30. The effect of induced mood on prices in experimental asset markets (2010)
  31. Emotions and chat in a financial markets experiment (2011)
  32. Yahoo Finance search and earnings announcements (2016)
  33. Can twitter posts predict stock behavior?: A study of stock market with twitter social emotion (2016)
  34. When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks (2011)
  35. Using millions of emoji occurrences to learn any-domain representations for detecting sentiment, emotion and sarcasm (2017) — link
  36. Empirical evaluation of gated recurrent neural networks on sequence modeling (2014) — link
  37. On persistence in mutual fund performance (1997)
  38. Ambiguity, information quality, and asset pricing (2008)
  39. Why don’t we agree? Evidence from a social network of investors (2020)
  40. Investor sentiment in the stock market (2007)
  41. The wisdom of crowds: Why the many are smarter than the few and how collective wisdom shapes business (2004)
  42. Results from a dozen years of election futures markets research (2008)
  43. The wisdom of Twitter crowds: Predicting stock market reactions to FOMC meetings via Twitter feeds (2016)
  44. Groups of diverse problem solvers can outperform groups of high-ability problem solvers (2004)
  45. Do analysts herd? An analysis of recommendations and market reactions (2010)
  46. Do analysts’ earnings forecasts incorporate information in prior stock price changes? (1991)
  47. Inefficiency in earnings forecasts: Experimental evidence of reactions to positive vs. negative information (2004)
  48. Herding among security analysts (2000)
  49. What makes markets predict well? Evidence from the Iowa Electronic Markets (1997)
  50. A short history of financial euphoria (1994)
  51. Is it the weather? (2008)
  52. Does the weather affect stock market volatility? (2010)
  53. Noise trader risk in financial markets (1990)
  54. Giving content to investor sentiment: The role of media in the stock market (2007)
  55. Bubbles, human judgment, and expert opinion (2002)
  56. From efficient markets theory to behavioral finance (2003)
  57. How emotions influence behavior in financial markets: a conceptual analysis and emotion-based account of buy-sell preferences (2020)
  58. Risk as feelings. (2001)
  59. A 12-point circumplex structure of core affect. (2011)
  60. Fast and accurate sentiment classification using an enhanced Naive Bayes model (2013)
  61. Glove: Global vectors for word representation (2014)
  62. Long short-term memory (1997)
  63. Bidirectional recurrent neural networks (1997)
  64. Framewise phoneme classification with bidirectional LSTM and other neural network architectures (2005)
  65. Estimating standard errors in finance panel data sets: Comparing approaches (2009)
  66. Twitter sentiment around the Earnings Announcement events (2017)
  67. k-means++: The advantages of careful seeding (2006)
  68. Software Framework for Topic Modelling with Large Corpora (2010)
  69. Snorkel: Rapid training data creation with weak supervision (2020)
  70. Predicting consumer default: A deep learning approach (2019)
  71. The Language of Earnings Announcements (2019)
  72. " I Just Like the Stock" versus" Fear and Loathing on Main Street": The Role of Reddit Sentiment in the GameStop Short Squeeze (2021)
  73. Investor emotions and earnings announcements (2021) — link
  74. Returns to buying winners and selling losers: Implications for stock market efficiency (1993)
  75. ‘Gut Feelings’ Are Driving the Markets (2020) — link
  76. How Emotion Hurts Stock Returns (2015) — link
  77. The behavior of individual investors (2013)
  78. Investor mood and financial markets (2010)
  79. Asset prices in an exchange economy (1978)

0 of 87 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.