1. Polymarket Brazil election odds put Bolsonaro near 80% before Lula won
Luiz Inácio Lula da Silva won Brazil's presidential runoff on Sunday, October 25, securing a fourth term hours after the Polymarket Brazil election market and its rival on Kalshi had priced his opponent, Flávio Bolsonaro, as the favorite at roughly 80%. Lula took 51.8% of valid votes to Bolsonaro's 48.2%, and traders are now arguing over whether the markets erred or simply lost a bet that was always going to fail one time in five.
A prediction market lets participants buy and sell contracts that pay out if a stated event occurs, and the trading price is commonly read as the crowd's estimate of the probability. By that reading, Bolsonaro was a heavy favorite on Sunday morning, and Lula stood at about 20%. The result therefore became an immediate test of how much confidence these venues deserve as forecasting tools.
The episode matters beyond Brazil because prediction markets have been promoted as faster and more reliable than opinion polls. A high-profile miss on a major national election invites a fresh look at that claim, while the defenders of the markets point out that an outcome priced at 20% is expected to happen regularly.
Calibration is the key concept in this dispute. A forecaster is well calibrated when events given a 20% chance happen about 20% of the time over many trials. Judged that way, the Brazilian result is not by itself a failure, yet it is the kind of outcome that prompts a closer look at how the price was formed, who was trading, and how much information the number really carried.
2. What were the official results and why do they matter?
Brazil's Superior Electoral Court counted 51.8% of valid votes for Lula and 48.2% for Bolsonaro. The Associated Press put Lula's winning margin at nearly 4.4 million votes. That is a clear gap in absolute terms, even if the percentage difference was modest.
Bolsonaro has refused to concede and said he will seek to have the result annulled. His argument rests on a late ruling by a Supreme Court justice that kept polling stations open for two additional hours. His challenge is pending, and its outcome has not been determined.
There is also a historical note. Lula finished second to Bolsonaro in the first round of voting, and the runoff reversed that order. It was the first time since direct elections returned in Brazil in 1989 that the candidate who came second in the first round went on to win the runoff, a fact that shows how unusual the path to victory was.
The dispute over the late ruling extending voting hours is likely to dominate the political aftermath. Whether any court acts on the annulment request will determine if the result is treated as settled or contested in the weeks before the inauguration, and it is also what keeps the election in the headlines for market watchers.
3. How did the odds move on Kalshi during election night?
Trader Michael McDonough described the swing on X. He wrote that Bolsonaro "traded at 80%+ for two days" before the count shifted the picture. According to his account, Lula's price then climbed from 20% to 70% while Bolsonaro dropped to 31%, and the move came on the heaviest trading volume the market had seen in its life.
Polymarket's Brazil election market attracted more than $249 million in total volume and has since resolved in Lula's favor. The size of that figure shows the contract was no niche product; large sums were staked on the outcome, which raises the stakes for any argument about the quality of its pricing.
One further detail adds to the picture: prediction markets are banned in Brazil. The country's voters and many of its residents therefore could not legally take part in a market devoted to their own election, leaving the price to be set largely by outside traders.
Heavy volume on a single night illustrates how these platforms operate during major events. Attention and money arrive in bursts, usually when the outcome is about to be revealed, and the biggest moves can happen in a short window. That concentration makes the final hours of trading far more informative about reaction speed than about long-run forecasting skill.
4. Were the markets out of line with the polls?
A tally of runoff polls compiled by the Rio Times gave Bolsonaro only about 51%, a far narrower advantage than the roughly 80% shown by the markets. A polling edge of that size would normally be described as close to a coin flip, which suggests the market prices reflected something other than the polling average.
The same analysis pointed to thin trading in Polymarket's order book. It estimated that about $30,000 of selling at 5:15 p.m. Eastern time would have shaved five points off Bolsonaro's price. In a market with that little depth on one side, a relatively small order can move the headline probability noticeably, making the displayed figure less sturdy than it looks.
The analysis also found that prices moved 30 to 40 minutes before the vote count became visible. Early movement of that kind may reflect traders who were watching results as they arrived before the wider public. It underlines that a market price is a record of what traders did, not an independent measurement of the truth.
Order-book depth describes how many buy and sell orders sit near the current price. When the book is shallow, a trader can shift the price with little capital, and observers may mistake that shift for a change in collective belief. Because the market was reported to have handled more than $249 million in total volume, the thinness at specific moments is a notable contrast, since total volume can hide the fact that only a small amount is available at the best prices at any one time.
5. Was it a failure or just a one-in-five outcome?
Reaction from commentators was split. Joe Weisenthal asked on X whether the result would count as "one of the biggest prediction market whiffs of all time." Harry Crane replied with a joke, asking whether a 4-1 underdog had ever won in the history of betting.
Nic Carter, who had pointed out earlier in the day that the market was "80/20 Bolso/Lula," later raised the other possibility. He asked whether the outcome was simply "normal variance that happens one in every five times." That is the statistical defense of the markets: a well-calibrated forecaster who assigns 20% to an event should see it happen in about one case out of five, so a single result cannot prove the pricing was wrong.
The difficulty is that one election offers a single data point. To judge whether a market is well calibrated, analysts need to look at many events and check whether outcomes priced at 20% occur roughly 20% of the time. A single surprise does not settle the matter in either direction, which is why the argument remains open.
The exchange also shows how social media has become part of the market conversation. Traders, analysts and commentators posted their reactions in real time, framing the result as either a miss or an expected tail event. Those framings spread quickly, and they influence how the wider audience, including regulators and potential users, comes to view prediction markets.
6. Why does the result raise questions about prediction markets as a source of truth?
The central issue is how much weight the public and the media should give to market probabilities. When a price is quoted as though it were a forecast, readers may treat it as more authoritative than a poll. Thin liquidity, trading ahead of public results, and the absence of local participants all complicate that picture.
Defenders argue that markets aggregate information quickly and put real money behind each view. Critics reply that the money can be concentrated in a few hands, that prices can be pushed by modest orders, and that a probability is not a prediction of what will happen. Both positions are reinforced by the Brazilian result, which supports the first view if one accepts the one-in-five defense and the second view if one focuses on the gap with the polls.
The outcome of this debate has practical consequences. Platforms such as Polymarket and Kalshi are competing for users and regulatory acceptance, and they rely on the argument that their prices are informative. A visible miss on an event of this size gives skeptics a ready example.
There is a further consideration for anyone reading a price as a probability. The number is shaped by the traders who choose to participate, by the fees and limits of the platform, and by the money available to them. In a country where prediction markets are banned, the pool of informed local participants may be smaller, which can leave the price more exposed to outside sentiment and less anchored to polling or on-the-ground knowledge.
7. What happens next?
Brazilian stocks sold off after the result. Lula is due to begin his new term on January 5, 2027, according to the Associated Press, while Bolsonaro's annulment challenge remains pending and is the next development to watch.
For prediction markets, the next tests will come with other major political contests. Whether traders and the platforms adjust how they handle thin order books and early price moves will shape how seriously their probabilities are taken the next time an election result diverges from the odds.
In the meantime, Lula's return to office and the unresolved legal fight over the vote will remain the dominant story in Brazil. The market debate is a side question for most Brazilians, but for the prediction market industry it is a live test of credibility. Each future election will give analysts another data point on whether the prices were calibrated, and repeated large misses would weigh more heavily on the case for these markets than a single upset.