Political Shocks and Price Discovery in Prediction Markets: Evidence from the 2024 U.S. Presidential Election

Abstract

Using transaction-level matched trades from Polymarket's 2024 U.S. presidential-election contracts, we study how prediction markets process major political shocks. We focus on three events with precise timestamps: the first Biden-Trump debate, the Trump assassination attempt, and Biden's drop out. We document large bursts of activity on both extensive and intensive margins, concentrated among high-intensity incumbents, and show that pre-event net exposure predicts abnormal post-event trading and position flips. To link order flow to prices, we estimate a Kyle-style price-impact measure and a Glosten-Harris decomposition that separates permanent from transitory order-flow effects, complemented by variance-ratio dynamics and a bounded two-sidedness index. Across shocks, price discovery differs sharply: the debate exhibits stronger transitory pressure and partial reversal, the assassination attempt features a more permanent repricing, and the drop out episode combines heavy trading with muted net price changes and high two-sidedness, consistent with disagreement under Knightian uncertainty.

Figures (13)

• Figure 1: Price and Transaction Volume of Different Candidate Markets
• Figure 2: Number of New Traders in Each 5-minute Interval Across Events
• Figure 3: Cumulative Abnormal Plots
• Figure 4: Price and Transaction Volume of Trump Tokens Around Events
• Figure 5: Event-time price impact and decomposition