When prediction markets disagree: cross-platform dispersion and equity volatility in the 2024 U.S. election

Official URL: http://dx.doi.org/10.1016/j.frl.2026.110664

We ask whether disagreement across prediction markets helps predict the volatility of politically sensitive equities during the 2024 U.S. presidential election. Using 5-minute odds from seven betting platforms and intraday prices for 71 politically exposed firms, we construct hourly measures of cross-platform dispersion and realized volatility. Higher disagreement predicts higher volatility in a partisan long–short portfolio: after controlling for broad realized and implied market volatility, a one-standard-deviation increase raises spread volatility by between 4% and 11%, depending on the intraday price source, with the range reflecting vendor differences in quote coverage and timestamp conventions. A formal selectivity test shows that the relationship is significantly stronger for politically exposed equities than for the broad market or a clean non-political (utilities and consumer-staples) benchmark. Consistent with a predominantly second-moment, uncertainty-based channel, disagreement predicts absolute and squared returns far more strongly than signed returns, and its effect persists for roughly one trading day. We interpret the evidence as selective predictive content specific to the 2024 episode rather than as a causal estimate or a universal pattern.