Brazil election result exposes prediction market mispricing

Brazil’s presidential runoff has ended with a result that prediction market traders were not expecting. Luiz Inácio Lula da Silva defeated Flávio Bolsonaro on Sunday, and he secured a historic fourth presidential term. This is despite entering election day as a clear underdog on Polymarket.

Let’s remember, as Predictbook recently reported, traders had Bolsonaro winning at a whopping 81% on October 15 and at 72% on October 22, just three days before the runoff.

This was despite the final surveys showing Bolsonaro with a narrow lead and often within the margin of error. Yet traders were pricing his chances so high that this competitive race had turned into what looked like a comfortable favorite.

It turns out, it was much closer than that, with Bolsonaro in a far more disadvantageous position than the traders had priced him.

When market confidence outruns the polls

The outcome shows a striking disconnect between polling and market pricing. It’s a firm reminder that prediction market prices shouldn’t be taken as voter intentions. Polls estimate support among respondents at a particular moment, and market prices reflect the expectations of people trading contracts.

Bolsonaro did outperform expectations in the first round with 47% of the votes compared to Lula’s 45. Still, according to Reuters, Lula ultimately won by more than four million votes. Left-wing President Lula ended the race with 51.79% of the votes, compared with Bolsonaro's 48.21%.

All this said, Lula’s victory can serve as a lesson to traders. It exposed the risk of treating traders’ conviction as a reliable substitute for electoral evidence that may point them to a more probable outcome. The overconfidence was costly for those who backed Bolsonaro at the prevailing odds.

The Brazil Presidential Election has seen nearly $250 million in volume, making it one of the most popular markets among traders.