Polymarket gave Luiz Inácio Lula da Silva only about a 20% chance of winning Brazil’s presidential runoff before voting began on Sunday. Lula nevertheless defeated right-wing challenger Flávio Bolsonaro, securing a fourth term and delivering a sharp reminder that prediction-market prices are probabilities—not guarantees.
The result has prompted debate over why the market leaned so heavily toward Bolsonaro and was slow to adjust as election-night information emerged. According to Decrypt, some traders blamed retail participants for keeping Bolsonaro’s implied odds artificially high. Other analysts argued that Brazil’s ban on prediction markets, limited liquidity and a lack of local participation were more important factors.
How the market missed the result
Bolsonaro led the first round of voting at the start of October but did not win the majority needed to avoid a runoff. Lula then won the second round with a majority, despite entering election day as a substantial underdog on Polymarket.
Polymarket contracts settle at $1 when the predicted event occurs. A contract trading at $0.60 therefore represents an implied probability of roughly 60%. Before the runoff, Lula’s contract was priced at about $0.20, while Bolsonaro’s stood near $0.80.
The market’s adjustment was also criticized during the vote count. Pseudonymous Polymarket trader FhantomBets told Decrypt that professional, high-stakes traders—often called “sharps”—believed Lula’s true chance was closer to 30% to 40% on election day. The trader said Lula’s “Yes” contract reached $0.60 at about 6:15 p.m. Eastern Time, when the result was already close to being decided, but that retail traders continued to offer substantial liquidity at prices implying Bolsonaro remained strongly favored.
FhantomBets said several large traders sold longer-term positions to free capital for the Brazil market. In that account, less experienced participants who opposed Lula helped preserve an inaccurate price until late in the count.
The role of regulation and local knowledge
Luke Nolan, senior researcher at CoinShares, challenged the idea that retail participation alone explained the error. He said additional users can provide liquidity and different perspectives, including local knowledge, and argued that the key question is whether participants bring information and can trade on it.
Nolan instead identified Brazil’s restrictions on prediction markets as a likely structural problem. Brazil imposed a broad ban on prediction-market platforms in April, citing investor-protection concerns and rising gambling addiction. Finance Minister Dario Durigan said the platforms violated betting rules and operated without sufficient regulatory oversight.
Einar Knobel, chief executive and co-founder of sportsbook and prediction-market firm TxODDS, said the market was shaped mainly by participants outside Brazil. He told Decrypt that traders focused on the first-round result, polling headlines and the broader narrative surrounding the election, rather than information from a deep domestic trading base.
Liquidity was also relatively limited. Knobel said the Brazil election market attracted only 4.5% of the volume generated by the 2024 U.S. presidential election. With fewer participants and little local knowledge represented in the order book, he said there was less capacity to correct an inaccurate price.
Nolan said traders may have placed too much weight on Bolsonaro’s first-round performance. They also may have underestimated the rise in turnout between the two rounds and made incorrect assumptions about how supporters of eliminated candidates would vote in the runoff.
A major upset, but not proof prediction markets are broken
The Brazilian election was a significant miss, but Nolan cautioned against treating one outcome as evidence that prediction markets have stopped working. A price implying an 80% probability still leaves a 20% chance that the event will not occur. Across a sufficiently large number of events, outcomes assigned a 20% chance should happen approximately one-fifth of the time.
Nolan also said prediction markets should be assessed against other forecasts available at the same time, rather than judged in isolation. He pointed to traditional betting markets as a comparison: Betfair gave Hillary Clinton an 83% chance of winning on the day before the 2016 U.S. election, although she ultimately lost.
The more serious concern, Nolan said, would be a recurring pattern of excessive confidence or a systematic failure to absorb relevant information. The Brazil result therefore raises questions about market design, liquidity and access to local information, while also illustrating the limits of interpreting any single probability as a certainty.
Moneyiar market picture
Moneyiar data recorded at the stated times on October 25, 2026, showed mixed performance among the listed digital assets:
- Lighter (LIT): $3.1819342396, down 1.45% over 24 hours and up 5.12% over seven days, as of 22:39 +03:30.
- edgeX (EDGE): $0.3244313833, up 0.81% over 24 hours and 13.63% over seven days, as of 22:38 +03:30.
- Comedian (BAN): $0.0575169485, down 4.44% over 24 hours and up 7.23% over seven days, as of 22:38 +03:30.
- Zilliqa (ZIL): $0.0018973706, down 5.04% over 24 hours and 14.28% over seven days, as of 22:39 +03:30.
These figures describe market performance at those specific times and do not provide a forecast of future prices.
