Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 26, 2026

Price Levels Are Not Enough: A Counterfactual Test for LLM Seller Collusion, with Negative Results for Price-Based and Probe-Based Detectors

Navien Potheri

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

Read the report

Report: Price Levels Are Not Enough: A Counterfactual Test for LLM Seller Collusion, with Negative Results for Price-Based and Probe-Based Detectors

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

LLM agents that set prices in a repeated market can settle above the competitive benchmark. We ask whether that can be detected from the market record. Using a continuous double auction with LLM sellers, we pre-registered and tested price-based labels, a price-free detector, a sequential live test, a monitoring notice, a disclosed cutoff, a linear probe on hidden states, and two market-design interventions. Most tests fail: labels move with the cutoff, detectors trained on one set of runs do not transfer, the live test does not separate collusive from competitive runs, and the probe's price signal is confounded with game position. One result is positive but preliminary. In the experimental design we vet GPT, open models and Claude. The Claude sellers, showed a unique property, prices above $90 fall by about 4.5 points when rivals' asks are hidden on the same seed, measured against a fresh-visible control, while GPT and open-weight sellers show no comparable effect at this design. We conclude that price levels alone cannot establish coordination, that a counterfactual that hides rivals can, and that the effect we observe is specific to one model family in our runs. The Claude result has not replicated on an unselected sample and needs a larger study

Reviews

Judging this Sprint?

Review “Price Levels Are Not Enough: A Counterfactual Test for LLM Seller Collusion, with Negative Results for Price-Based and Probe-Based Detectors”

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. This submission attempts to build collision detectors that use only price levels submitted by the agents (not, e.g., their internal state or chain of thought).

    They test several reasonable designs and find that none of them work reliably. This shows that building such a detector is difficult.

    Overall, the paper is clear and well presented.

Cite this project

@misc{potheri2026price,
  title = {{Price Levels Are Not Enough: A Counterfactual Test for LLM Seller Collusion, with Negative Results for Price-Based and Probe-Based Detectors}},
  author = {Navien Potheri},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/price-levels-are-not-enough-a-counterfactual-test-for-llm-seller-collusion-with-negative-results-for-pricebased-and-probebased-detectors-kskx}},
  url = {https://apartresearch.com/sprints/projects/price-levels-are-not-enough-a-counterfactual-test-for-llm-seller-collusion-with-negative-results-for-pricebased-and-probebased-detectors-kskx}
}

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