Track or Follow? A Jev-Compatible Multi-Agent Trading System When Web News Lags the Market

Paper: arXiv 2611.10414

Authors: Yebo Feng, Guangyan Gan, Liangrui Pan, Ruichao Liang, Cong Wu, Jiahua Xu, Yang Liu

Abstract

LLM agents increasingly trade on the Web news they read, yet much Web news about listed firms is published after, and about, the price move it reports, so a reader can agree with the market by following it instead of reading the event. We ask whether LLM judgments of Web news track the market’s reaction to the reported event or follow it. To answer, we built a multi-agent trading system whose agents share only point-in-time stores and answer typed questions through the interface of Jev, a hosted typed decision model; code sets every position under a proved bound on how far text can move it. In a hash-committed test of a Jev-compatible reader built from a pinned LLM’s option probabilities, item judgments track announcement-window abnormal returns (weekly Spearman 0.119), also after the LLM’s release, and event types agree with the issuers’ 8-K filings (AUC 0.833); Jev’s judgments of news published after its own release track the reaction as well (descriptive). Exploratory analyses show where the association comes from: 72% of earnings news items appear after the issuer’s filing, contributor-written items often report the price move, and from the first trade after publication the association falls to 0.023. At a weekly horizon, the confirmatory test detects no rank information over code-only features. Under attack, decision cards traced 88% of successful attacks to the planted item. We propose that market-graded tests of Web-text readers report the criterion from publication, not only from the event.

Complexity vs Empirical Score

• Math Complexity: 6.5/10
• Empirical Rigor: 8.0/10
• Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents a sophisticated multi-agent trading system with a strong emphasis on empirical validation and rigorous testing of LLM judgments against market reactions. Its novel approach to mitigating look-ahead bias and evaluating LLM performance in a trading context is commendable.

Research Flowchart

  flowchart TD
    A[Research Goal: Do LLM Judgments of Web News Track or Follow Market Reactions?] --> B{Methodology: Multi-Agent Trading System with Jev-Compatible LLM Reader};
    B --> C[Data/Inputs: Web News, Issuer 8-K Filings, Market Price Data (Abnormal Returns)];
    C --> D{Computational Process: Jev-Compatible Reader (Pinned LLM's Option Probabilities), Code-Only Features, Decision Cards};
    D --> E[Key Finding 1: LLM Item Judgments Track Announcement-Window Abnormal Returns (Spearman 0.119 Weekly)];
    E --> F[Key Finding 2: Event Types Agree with 8-K Filings (AUC 0.833)];
    F --> G[Key Finding 3: 72% Earnings News After Filing, Weak Association from First Trade (0.023)];
    G --> H[Outcome: Propose Market-Graded Tests Report Criterion from Publication, Not Only Event];