Preference Optimization Drives Monoculture in LLM Prediction Markets

Summary

Directly tests whether the error-independence that prediction markets rest on survives when the crowd is LLM agents, and finds it does not: DPO-tuned agents show pairwise error correlation rho = 0.70, reducing ten agents to roughly 1.4 effective independent forecasters, with N_eff flat from N=5 to N=40 and the 10-agent market (67.6%) failing to beat a single standalone agent (70.2%). Two ablations isolate preference optimization as the causal driver, replicated across labs and scales, and cross-model diversity is the strongest mitigation (rho 0.68 to 0.40). The sharpest available quantification of the correlated-error ceiling on market-based and confidence-weighted aggregation in an LLM population.

Contribution

Queued — not yet read.

Method

Queued — not yet read.

Relevance to RISE

Surfaced by the 2026-08-29 literature sweep as a high-priority candidate for the Architecture as Epistemology (ISR) paper (angle: correlated errors / homogeneity across LLM agents; LLM agents in forecasting; aggregating LLM judgments). Verified against: Opened https://arxiv.org/abs/2606.26583 (arXiv abstract page, v1 25 Jun 2026)

Critique / open questions

Queued — not yet assessed.

Key quotes

Prediction markets rest on the independence of participant errors. (abstract)

As LLM agents become active traders on platforms like Kalshi and Polymarket, we ask: does this independence hold when the crowd is composed of LLMs? (abstract)