Record prediction-market turnover contrasts with setbacks for heavily favored candidates

Prediction markets recorded a new monthly high of 21 billion dollars in October volume, according to WiseVoter. That activity coincided with high-profile forecasting misses in Brazil, Serbia and Wisconsin.

Strong favorites, divergent results

In Brazil, Polymarket assigned Flávio Bolsonaro about 80 percent odds in the October 25 presidential runoff, according to WiseVoter. Luiz Inácio Lula da Silva won instead.

In Wisconsin, both major platforms put Francesca Hong at about 95 percent in the Democratic governor primary. WiseVoter says she lost by less than half a point—a narrow electoral margin despite the strong market confidence in her winning.

In Serbia, Polymarket gave Aleksandar Vučić a 70 percent probability, while his eventual vote share was 39.2 percent, according to WiseVoter. Those figures measure different things: an outcome’s implied probability is not a forecast of a candidate’s share of the vote.

What it means for prediction markets

• Record turnover measures trading activity, not forecasting accuracy. October’s volume milestone and the cited electoral results should be assessed separately.
• The Brazil and Wisconsin outcomes illustrate that even heavily favored candidates can lose. These individual results alone do not establish whether market probabilities are systematically miscalibrated.
• The Serbia comparison requires particular care: assessing a market’s performance requires its precise outcome definition, rather than a direct comparison between probability and vote share.