ABSTRACT This paper examines whether prediction market data can forecast sovereign credit risk during periods of geopolitical conflict. Using prices from 152 political event contracts traded on Polymarket, we construct a market‐based measure of geopolitical uncertainty and aggregate information using a neural network. We evaluate its ability to predict daily changes in Ukrainian sovereign bond spreads from September 2024 to September 2025. Models incorporating the prediction market signal outperform standard benchmarks in out‐of‐sample tests, with forecast improvements of up to 4.6% for the 5‐year spread. These results indicate that prediction markets capture geopolitical risk not reflected in traditional financial variables.
Becker et al. (Thu,) studied this question.