Paper: arXiv 2604.21433

Authors: Sebastian Lehner, Alejandro Lopez-Lira

Abstract

Frozen large language model (LLM) checkpoints extract information from pre-cutoff public text that is associated with future fundamentals and equity returns beyond standard contemporaneous valuation measures. Because each frozen checkpoint has a fixed knowledge cutoff, it can be interpreted as a compressed representation of publicly available textual information at a given point in time. We treat twelve OpenAI snapshots spanning 2021-2025 as time-stamped summaries of the public textual record and extract a sector-neutral LLM outlook score for roughly 7,000 U.S. equities per cross-section. The outlook score is positively associated with analyst revisions, target-price changes and one-month cross-sectional returns in both Fama-MacBeth regressions and pooled panels with model fixed effects (t = 6.02), after direct controls for market-implied valuation and standard factors. Predictability broadly increases with the return horizon, despite a non-monotonic intermediate dip, and, in the pooled panel, is stronger for firms with high analyst coverage, consistent with the view that the bottleneck is not investor inattention but the cost of aggregating dispersed qualitative information across many documents.

Complexity vs Empirical Score

  • Math Complexity: 3.5/10
  • Empirical Rigor: 6.0/10 (capped from 8.0: transaction costs not modelled; see penalty flags)
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper presents an innovative conceptual framework using historical LLM snapshots as time-stamped textual aggregators to test market efficiency and price discovery. It features strong empirical execution using rigorous Fama-MacBeth regressions and pooled panels across 7,000 equities, relying on applied econometrics rather than heavy mathematical derivations.

Research Flowchart

  flowchart TD
    A["Research Goal: Test if frozen LLM checkpoints contain unpriced qualitative info predicting stock returns"] --> B["Inputs: 12 OpenAI snapshots (2021-2025) & ~7,000 US Equities"]
    B --> C["Extract Sector-Neutral LLM Outlook Scores from Pre-Cutoff Text"]
    C --> D["Statistical Testing: Fama-MacBeth & Pooled Panel Regressions with Factor Controls"]
    D --> E["Outcome: Scores predict 1-month returns (t=6.02), analyst revisions & target prices"]
    D --> F["Outcome: Predictability increases over longer horizons & is stronger in high-coverage firms"]