Can ChatGPT Compute Trustworthy Sentiment Scores from Bloomberg Market Wraps?

ArXiv ID: 2401.05447 “View on arXiv”

Authors: Unknown

Abstract

We used a dataset of daily Bloomberg Financial Market Summaries from 2010 to 2023, reposted on large financial media, to determine how global news headlines may affect stock market movements using ChatGPT and a two-stage prompt approach. We document a statistically significant positive correlation between the sentiment score and future equity market returns over short to medium term, which reverts to a negative correlation over longer horizons. Validation of this correlation pattern across multiple equity markets indicates its robustness across equity regions and resilience to non-linearity, evidenced by comparison of Pearson and Spearman correlations. Finally, we provide an estimate of the optimal horizon that strikes a balance between reactivity to new information and correlation.

Keywords: Sentiment Analysis, Large Language Models, ChatGPT, Market Returns, News Impact, Equity

Complexity vs Empirical Score

• Math Complexity: 2.5/10
• Empirical Rigor: 7.0/10
• Quadrant: Street Traders
• Why: The paper employs basic statistical formulas and correlation analysis but lacks advanced mathematical derivations, while its strong empirical rigor is evidenced by a long-term dataset, cross-market validation, and implementation-heavy testing with real-world financial news.

  flowchart TD
    A["Research Goal<br/>Can ChatGPT compute trustworthy<br/>sentiment scores from Bloomberg Market Wraps?"] --> B
    subgraph B ["Data & Methodology"]
        B1["Bloomberg Market Summaries<br/>2010-2023"] --> B2["Two-Stage Prompt Approach<br/>(ChatGPT Sentiment Scoring)"]
        B2 --> B3["Correlation Analysis<br/>Pearson & Spearman"]
    end
    B3 --> C{"Key Findings"}
    C --> C1["Significant Positive Correlation<br/>Sentiment → Short/Medium-term Returns"]
    C --> C2["Correlation Reverts<br/>Negative over Long Horizons"]
    C --> C3["Robust Across Markets<br/>Resilient to Non-linearity"]
    C --> C4["Optimal Horizon Estimation<br/>Balances Reactivity & Correlation"]