Prediction markets have earned a reputation for accuracy, often outperforming traditional polls and expert forecasts. But when they fail, they fail spectacularly. Understanding what is a prediction market and its mechanics is crucial, yet even more valuable is studying its blind spots. This article examines the most notorious prediction market failures to reveal what prediction market basics miss and why collective intelligence sometimes falls flat.
Brexit 2016 and the deep-money fail
The 2016 Brexit referendum shocked prediction markets worldwide. Platforms like Betfair and PredictIt showed Remain winning with 70 to 80 percent probability just hours before polls closed. The prediction market definition centers on aggregating information through real money bets, yet deep-pocketed traders flooded markets with Remain contracts based on outdated polling models.
Markets like Polymarket and Kalshi did not exist yet, but the failure taught a hard lesson. How prediction markets work depends on diverse, independent information sources. When wealthy traders dominate liquidity and rely on the same flawed polls, prediction market mechanics break down. The wisdom of crowds prediction markets promise evaporates when the crowd becomes an echo chamber.
2016 US election: markets and polls both off
Prediction markets gave Hillary Clinton an 80 to 90 percent chance of victory on election night 2016. Both prediction markets vs polls comparisons failed because they shared the same blind spot. Polling vs prediction markets debates assumed markets corrected poll bias, but traders simply bet on poll averages.
The Iowa Electronic Markets and Hollywood Stock Exchange, pioneers in prediction market history, also missed the outcome. Types of prediction markets, whether binary markets vs scalar markets or categorical prediction markets, all struggled. The issue was not market design but information quality. Collective intelligence forecasting only works when participants have access to genuine signals, not recycled narratives.
Low-liquidity markets that drifted from reality
Smaller prediction markets often drift into fantasy pricing when liquidity dries up. A single trader with $500 can move a thinly traded contract by 20 percentage points. Binary contracts explained in textbooks assume continuous trading and price discovery, but real platforms see days without activity.
Crowd accuracy research shows that prediction markets need at least dozens of active traders to self-correct. Below that threshold, prices reflect the last trader’s hunch, not collective wisdom. Platforms like Kalshi and Polymarket address this by focusing on high-interest events, but niche markets remain vulnerable.
What failure modes teach platform designers
The structural reasons markets miss
Prediction market basics assume rational actors and efficient information flow. Reality is messier. Herding behavior, where traders follow the crowd rather than independent analysis, amplifies errors. Prediction market mechanics also struggle with correlated information, when all participants read the same news sources or trust the same experts.