The Conversation Turns First: Crowd Discussion and Price Reversals in Prediction Markets

2026-09-23

AI Summary

This study investigates the predictive capacity of comment signals in prediction markets for forecasting trading behavior and outcome reversals. Leveraging Polymarket data, the research integrates correlation scanning, six-category classification experiments, and NLP-based sentiment-stance analysis to demonstrate that attention approximates modeled trading histories, with systematic evaluation conducted via PR-AUC and ROC-AUC metrics. The findings reveal that toxicity indicators significantly outperform random baselines in predicting buy-sell directions, while multi-feature fusion improves ROC-AUC to 0.788. This work confirms a shared predictive linkage between discussion information and trading responses, underscoring the core value of textual signals in short-term prediction markets.

Abstract

Prediction markets combine trading with public discussion of the same events. We examine whether comment-derived signals predict subsequent activity, buying direction, and changes in the leading outcome. A correlation sweep across 79 non-political Polymarket markets guides six classification experiments comparing comment features, trading features, and their combinations. On live blocks containing comments, attention nearly matches trading history in predicting heavy trading within 18 hours (PR-AUC 0.786 versus 0.790, against prevalence 0.606), with its relative advantage concentrated in short markets. Comment content carries directional information: toxicity ranks future buying direction above chance in 43 of 53 scored markets, while adding attention, sentiment, and stance to flow history increases ROC-AUC from 0.780 to 0.788. Leadership changes are predicted primarily by market state. Stance shifts against the leader before reversals in 23 of 27 evaluable markets, but adds no clear improvement in individual-block forecasting. These results distinguish attention from directional support and price uncertainty. They establish predictive associations consistent with discussion and trading responding to shared information, without identifying a causal effect of comments on markets.

Problem

• prediction markets
• crowd discussion
• price reversals
• comment signals
• trading activity

Innovation

• Prediction Markets
• Crowd Discussion
• Price Reversals
• Text Features
• Classification Experiments