With Brazil’s presidential runoff five days away, prediction markets are assigning Flávio Bolsonaro roughly four chances in five to defeat Luiz Inácio Lula da Silva, while polling points to a near tie. The gap is not evidence that one side has sampled voters incorrectly: Polymarket’s price reflects trading in a contract that pays on the winner, while a poll estimates stated vote intentions. In a 21 October snapshot, Polymarket’s election market showed Bolsonaro around 80% and Lula around 20%; Kalshi’s live market put Bolsonaro near 79% and Lula near 21%. A live market tracker recorded about $629,000 and $605,000 in 24-hour trading for the respective Polymarket outcomes. These are bets and market prices, not polls, and can change quickly.
The latest comparison is striking because the polling margin is only a few percentage points. Rio Times’ tracker showed a 16-poll average of about 51% for Bolsonaro to 49% for Lula on 20 October, with the exact average shifting as new surveys arrived. Individual results ranged from Real Time Big Data’s 52-48 Bolsonaro lead to Data Ranking’s 50.4-49.6 edge for Lula. The tracker’s simple mean gives each poll equal weight rather than modeling pollster quality or field dates.
A Narrow Vote Margin Can Produce a Wide Win Probability
Vote share and probability of winning answer different questions. A survey measures stated vote intention at a moment; a 51-49 split is not a forecast that the leader wins by exactly two points. A market contract priced at 80 cents implies traders are willing to buy a dollar payout for about 80 cents if Bolsonaro wins, subject to spreads, fees and liquidity. It is an implied probability, not 80% of voters backing him.
The difference can arise because market participants translate polling into a distribution of possible outcomes. If they believe polls are noisy, turnout uncertain or late movement likely, they can still conclude that one candidate has a high chance of finishing first even when the central polling estimate is close. Recent momentum, regional vote patterns and assumptions about undecided voters may also shape bets. Participants are not a representative sample of voters; their information, incentives and risk limits vary, and large trades can move prices.
Market depth also matters. Polymarket’s election contract has accumulated substantial lifetime trading, but cumulative volume is not the same as money committed at today’s price, and turnover can count repeated trades of the same capital. The two current outcome figures are market turnover, not cash staked or a measure of voter support.
First-Round Signal and Limits
The first round offers a useful but incomplete check. On 4 October, Bolsonaro received 47.03% of valid votes and Lula 45.16%, a 1.87-point gap, according to the official tally reported by NBC News. Bolsonaro advanced in front after many final polls had Lula leading or the race close. That outcome showed that the polling picture understated Bolsonaro’s first-round support, while the result also confirmed that neither candidate had a majority. It does not prove that an 80% market probability is calibrated correctly: a first-round vote share is not a head-to-head runoff result, where transfers, abstentions and campaign shifts matter.
Markets did capture a possibility that the polls’ close estimates did not resolve: Bolsonaro could emerge with momentum and a narrow lead. But confidence in a win can overshoot the information contained in a small polling margin. The useful comparison is how each method handles uncertainty, new information and turnout.
What a Bolsonaro or Lula Win Could Mean for Assets
Investors have treated a Bolsonaro victory as potentially friendlier to fiscal adjustment and private-sector policy, helping Brazilian equities and the real after his first-round performance. Reuters reported that the Bovespa hit a record after the first-round result and that the real posted its strongest one-day gain in four years. Those moves show an election premium in prices, but a tight result, contested count or policy disappointment could reverse positioning. The US debate over prediction-market regulation is separate from Brazil’s electoral rules. Neither candidate’s victory probability predicts crypto policy; investors should assess campaign proposals and official regulation rather than infer a rules change from election odds.
A Lula win could unwind some of the trades tied to expectations of political change, although currency and equity moves would also depend on fiscal signals, rates, commodities and global risk appetite. Neither candidate’s victory probability predicts rules for digital assets; investors should watch campaign proposals and post-election appointments.
What to Watch on Sunday
Three indicators will clarify the gap: the final Datafolha and Quaest polls, turnout and regional margins as results arrive, and whether market prices move on new information or thin liquidity. Until ballots are counted, Polymarket and Kalshi are useful measures of trader expectations, not substitutes for Brazil’s polls or election authority.