Prediction Markets Keep Missing Elections, and the Midterms Are Next

Prediction markets missed another run of high-profile elections, and the misses landed weeks before the U.S. midterms. The Wall Street Journal reported on October 31, 2026 that markets on elections in Brazil, Serbia and Wisconsin were upset by candidates the markets had treated as long shots. Polling in those races had put one candidate as high as 90%, and the polling data itself remains in dispute.

Brazil is the clearest case. The New York Times reported on October 30 that Polymarket favored Flávio Bolsonaro going into Election Day in Brazil's presidential contest, and he lost. Reuters reported that Lula was re-elected, reversing market bets on a Bolsonaro win.

Kalshi has published its own account of the Brazil outcome. By that count, its market missed the final margin by about 7 points, while the polls missed by about 5.6 points, as Yahoo News reported on October 31. On those figures, the polls landed closer to the result than the market did.

That comparison covers one race, so it cannot settle whether markets or polls are the better tool in general. It does matter for the midterms. A market price is usually built on top of polls and other inputs. If those inputs are wrong, the price inherits the error, and a fast-moving market can make a weak input look like a settled view.

Betting on U.S. midterm elections surged in 2026 on Polymarket and Kalshi, according to Reuters, and both platforms rose to prominence in 2024. Reuters also reported in September that researchers said prediction markets tracking the 2026 U.S. congressional elections are "extremely vulnerable" to manipulation, and that forecasts for those races can shift on small bets, according to Reuters.

In my view, the thing to watch is depth, not total volume. A market can be large in aggregate and still be thin in the specific contract a headline quotes. A thin contract can be pushed by a modest amount of money, and that is the number that usually ends up in the headline.

Kalshi says it resumed taking bets on U.S. election outcomes after a federal appeals court lifted a regulatory freeze, according to Kalshi's press page. The contracts traders can see on election outcomes are the same ones that moved during the Brazil and midterm coverage.

For a reader who follows these markets, the main lesson is plain. A prediction market price is a probability that traders have set, not a forecast that has been checked against outcomes. Its record in a handful of races, Brazil included, is a reason to treat it as one signal among several. The midterms will give us a larger sample. The number to track is whether the markets and the polls converge, and by how much each one misses. Anyone using these prices to make decisions should check a contract's liquidity, meaning how much money actually trades in it, and where its inputs come from before reading the headline percentage as a probability.