Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 25, 2026Hong Kong, Online

Detecting Coordination Risk among AI Trading Agents: Integrating Communication and Behavioral Evidence for Financial Auditing

XIE Tengda, DU Yunhao · Team AI Auditing

Submitted to AI Collusion Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Detecting Coordination Risk among AI Trading Agents: Integrating Communication and Behavioral Evidence for Financial Auditing

Open in a new tab (opens in new tab)
Code (opens in new tab)
Share

This study develops and evaluates an audit-oriented framework for detecting coordination risk among artificial-intelligence trading agents. The framework integrates non-semantic communication structure with behavioral alignment residualised against observable market-wide and common-shock conditions. The experiment uses a rule-based synthetic data-generating process rather than a reinforcement-learning model: coordinated sets, activity periods, and labels are specified for evaluation. Four scenarios represent independent activity, explicit coordination, weak and intermittent coordination, and a no-coordination common-shock counterfactual. Across 120 runs, five agents, and 500 periods per run, the integrated logistic model achieved a mean ROC AUC of 0.852 and an F1 score of 0.684 under repeated run-grouped holdouts. A common-shock ablation reduced ROC AUC to 0.737 and F1 to 0.584, showing that raw synchrony alone is not an adequate audit signal. A nested threshold procedure targeting a 10% G4 false-positive rate produced a mean out-of-sample G4 rate of 0.092. Semi-synthetic analysis using public daily Fama-French factor backgrounds yielded an ROC AUC of 0.844 and an F1 score of 0.697, but it does not constitute real-world validation. The proposed output is therefore a transparent referral score for human review rather than a finding of unlawful conduct.

Reviews

Judging this Sprint?

Review “Detecting Coordination Risk among AI Trading Agents: Integrating Communication and Behavioral Evidence for Financial Auditing”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

How much would this matter for the field if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. A clean, reproducible ML workflow: run-grouped splits, nested threshold selection, calibration metrics and a common-shock counterfactual. I ran the scripts and the performance and threshold tables reproduced exactly. The main limitation is that the "AI trading agents" are rule-based, not learning or LLM agents, so the study measures how well a classifier recovers signals the simulator plants. For example, G2 coordinating agents get extra within-set messages, which the message-rate features can pick up, and the "shock-controlled" alignment feature is residualised on the simulator's exact common-shock series, which a real auditor could only estimate. To strengthen it: use LLM or RL agents, estimate the common factor from observable market data, and test against coordination mechanisms the features weren't designed around. The report is 13 pages and its first 8 include a title page and table of contents; shortening the literature review would bring the results forward.

    Read full reviewShow less

Cite this project

@misc{tengda2026detecting,
  title = {{Detecting Coordination Risk among AI Trading Agents: Integrating Communication and Behavioral Evidence for Financial Auditing}},
  author = {XIE Tengda and DU Yunhao},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/detecting-coordination-risk-among-ai-trading-agents-integrating-communication-and-behavioral-evidence-for-financial-auditing-gvyx}},
  url = {https://apartresearch.com/sprints/projects/detecting-coordination-risk-among-ai-trading-agents-integrating-communication-and-behavioral-evidence-for-financial-auditing-gvyx}
}

More from AI Collusion Research Sprint

All 392 entriesBrowse all projects

A Sprint is running now

AI x Epistemics Research Sprint · Nov 13 - 15, 2026