Judith Levine says Polymarket midterms betting is being sold as crowd wisdom when a tiny slice of traders does most of the work. In her Guardian column, she argues that voters can be misled if they treat those prices as broad public judgment.
That challenge lands at a moment when prediction markets are already being discussed as election forecasters. The contrast is simple: most users provide volume, but the useful information appears to come from a small, skilled minority.
Judith Levine and Polymarket
Levine wrote that sites like Kalshi and Polymarket offer faux populism as elites reap the benefits and voters risk being misled. Her target is not betting itself, but the idea that the average account is doing the forecasting work.
Kalshi’s spokesperson gave a different account, telling: “This data is bottom-up, not elites-driven,” and adding, “Prediction markets are a way to see what people are saying when there’s money on the line.” That framing presents the platforms as a collective signal, not a specialized trading desk.
London Business School and Yale University
A 2025 study from Vanderbilt University found Polymarket was more accurate than polls in the 2024 elections, especially in the swing states. A working paper from scholars at London Business School and Yale University narrowed that claim sharply, saying “accuracy comes from a minority of persistently skilled traders, around 3% of accounts.”
That is the practical distinction for readers watching election prices: the market can still be informative even if most participants are not. The crowd may supply the flow, but the paper says the signal comes from a small group that keeps getting the pricing right.
The analysis
A separate analysis of 1.6 million Polymarket accounts traded since November 2022 found a sharper divide. It said casual traders are losing money while a small number of sophisticated pros are winning, with three-quarters of the profits going to 0.1% of the accounts.
The same analysis said at least 2.3 million accounts were on the site, but just 2,000 netted almost half a billion dollars. It also said the pros react quickly to public and financial news, gather poll data, buy access to big data streams that can cost $200,000 annually, and use algorithms to predict price movements and manage risk.
Pew Research said the median Polymarket user placed 46 trades across 10 active trading days in 2026. For casual traders, that leaves a market that can look democratic from the outside while functioning, in practice, like a highly concentrated information business.
The open question for Polymarket midterms is whether the platform’s apparent edge can hold without depending on that small class of persistent traders. If the next round of election betting repeats the same pattern, the market may keep being useful — but not for the reason its public image suggests.
