quant research // summarizing the paperLOADING
Hoyoung Lee, Suhwan Park, Seunghan Lee, Jun Seo, Jaehoon Lee, Sungdong Yoo, Minjae Kim, CheolWon Na +10 · 2026-06-28
The paper studies how large language models summarize financial documents like filings and earnings-call transcripts, and shows these summaries can be fluent and factually accurate yet still change the investment decision the original source would support. It identifies two failure modes decontextualization (keeping facts but stripping caveats) and model dependency (different summarizers giving different pictures), and proposes generating multiple summaries and auditing their disagreements against the source.
Why it matters: Anyone using LLMs to digest large volumes of financial disclosures should be aware that a plausible-sounding summary may quietly flip the conclusion the raw document supports. The proposed multi-summary auditing approach is a practical check for teams building AI-assisted research or agentic workflows.
⚠ This is a diagnostic/methodological study of LLM behavior on financial text, not a validated trading strategy, and it reports qualitative failure patterns rather than live decision outcomes.
Financial decision-makers face more information than they can directly inspect, making context compression necessary. Yet when large language models (LLMs) compress financial source material, they can alter the investment judgment supported by the original source. We frame this problem as information fidelity: compression loses fidelity when it changes the decision induced by the source. In agentic systems, such losses may recur across intermediate steps and amplify throughout the decision process. Across financial filings and earnings-call transcripts, we find that LLM-based compression can produce fluent and factually plausible compressed contexts that nevertheless alter downstream decisions. We analyze two diagnostic patterns associated with fidelity loss: decontextualization, where salient evidence is retained but separated from the caveats and contextual qualifiers needed for correct interpretation, and model dependency, where different compressors expose different views of the same source. We then propose Agentic Context Compression, which generates multiple candidate compressions and audits their disagreements against the original source. Our results suggest that financial compression should be evaluated not only by efficiency or factuality, but also by its ability to preserve decision-relevant context.
Go deeper: a full research-committee breakdown of this paper, its assumptions and failure modes, and how its method would apply to a specific ticker or your watchlist. See StockTools AI →
AI summary generated from the paper’s public abstract via arXiv; it may miss nuance — read the source before relying on it. Thank you to arXiv for its open-access interoperability; StockTools is not affiliated with arXiv, and all rights remain with the authors. Educational only, not financial advice.