StakeBench: Evaluating Language Understanding Grounded in Market Commitment

2026-05-25

AI Summary

This work proposes StakeBench, a novel evaluation framework for financial NLP that replaces conventional human-annotated labels with real market behaviors—such as positions, trading actions, and odds trajectories—to assess models’ understanding of market participants’ actual commitments. The framework introduces four diagnostic tasks: stance identification, action prediction, position-side recognition, and collective odds projection. Evaluated on 560,000 comments from Polymarket and Manifold paired with corresponding market records, the study reveals that 15 prominent large language models achieve only modest performance in stance identification (accuracy: 0.506–0.599), generally degrade to minority-class baselines in action prediction, and fail to consistently outperform a naive odds-based benchmark. By introducing “commitment-aware” metrics and clarifying the boundaries among commitment signals, latent beliefs, and causal effects, this work establishes a behaviorally grounded paradigm for financial language understanding.

Abstract

Existing financial NLP benchmarks often rely on labels supplied by outside observers, measuring how language is perceived rather than what speakers have committed to in the market. We introduce StakeBench, an evaluation framework for language understanding grounded in market commitment. StakeBench links 560,876 comments from 2,261 resolved markets to verified position, action, and market-odds records across Polymarket and Manifold. Supervision is derived from observable market behavior. Position sides, post-comment trading actions, and market-odds trajectories replace human annotation. Four diagnostic tasks test whether models detect market commitment, identify the revealed side, anticipate future action, and perform collective odds projection. Three commitment-aware metrics measure alignment with revealed preferences rather than perceived sentiment. Validity audits and explicit interpretation boundaries help distinguish observable commitment signals from latent belief and causal market-odds impact. Across 15 LLMs and 18 topics and platform settings, models partially recover position-side signals, with Directed Accuracy from 0.506 to 0.599, but show structural failures on later tasks. Ten of the fifteen models collapse to one or two action labels in future action anticipation, and no model consistently improves on the naive odds-direction baseline in collective odds projection. Model scale is not correlated with performance, finance-domain tuning does not improve revealed-side identification, and platform incentives strongly shape higher-order results. StakeBench is packaged with evaluation code and dataset under CC-BY 4.0.

Problem

• market commitment
• financial NLP
• language understanding
• revealed preference
• benchmark

Innovation

• market commitment
• revealed preference
• financial NLP benchmark
• observable behavior supervision
• collective odds projection