Prediction markets showed no significant partisan or demographic bias in election pricing, study finds

An NBER working paper by Prof. Zitzewitz, drawing on over 100 years of prediction market data, found no dimension of political affiliation, gender, race, or age where returns differed from zero beyond the standard error.

The only exception was non-US elections, where markets appeared to overrate right-leaning candidates, but that bias was not statistically significant. The findings are preliminary.