Brazil's first round changed the evidence. Now the first post-election polls will test whether the market's repricing was justified.
Brazil's presidential election has entered its decisive second round.
On October 4, Flávio Bolsonaro received 47.03% of valid votes, compared with 45.16% for Luiz Inácio Lula da Silva.
The difference was just 1.87 percentage points.
Yet the result was enough to challenge the assumptions behind several pre-election forecasts, including our own.
At ProbGap, we believe that new evidence should change forecasts when the evidence warrants it.
But the size of that adjustment matters just as much.
What the first round actually told us
Three questions now deserve particular attention.
1. How much of the polling miss was genuine late movement?
The final surveys underestimated Bolsonaro's first-round support. But analysts have offered different explanations, including tactical voting and turnout differences.
2. How will the remaining electorate behave?
The candidates who failed to reach the runoff collectively received approximately 8% of valid votes. Their voters do not necessarily transfer uniformly to either finalist.
3. What will the first post-election polls show?
Datafolha's new survey interviewed 2,520 voters between October 6 and October 8.
It provides the first major opportunity to compare post-election voter intentions with the assumptions made immediately after the first round.
That distinction is crucial.
A prediction market reacts immediately. Survey evidence takes longer to arrive.
The question is whether the two signals will converge.