October 26, 2026. Prediction markets have become the default price discovery engine for political, economic, and sports outcomes, and in most years they outperform polls, pundits, and prop desks. But the record is not perfect. From Brexit in 2016 to the 2024 Iowa caucuses, there is a short but consequential list of moments where the crowd was decisively wrong, and each one offers a lesson for how to read today’s markets. With US midterm control pricing on Kalshi tightening to 54% GOP Senate hold and Polymarket’s 2028 presidential primary contracts seeing record volume this week, it is a useful moment to revisit the upsets.
Brexit, 2016: The 88% Favorite That Lost
The clearest modern prediction market miss remains the June 2016 Brexit referendum. On the morning of the vote, Betfair, PredictIt, and the sterling-denominated bookmaker markets priced “Remain” at roughly 88% to 92%. Volume was heavy, liquidity was deep, and the number barely moved all day. By the time Sunderland reported, Remain was already a coin flip, and by dawn it was over. The miss was not a small one. A near-9x favorite lost outright, and the pound dropped more than 8% in a few hours as leveraged traders unwound. The lesson that has echoed through every subsequent market cycle: when a single binary event sits on top of a polarized electorate with a non-trivial turnout differential, the market’s confidence is often a function of trader geography rather than genuine informational advantage.
Trump 2016: The 85% Clinton Market
Five months later, the same pattern repeated at larger scale. Across PredictIt, Betfair, and the IEM, Hillary Clinton was pricing between 82% and 85% on election morning. By 9pm Eastern, as Florida and North Carolina moved, the market cratered 60 points in under three hours, one of the largest single-session repricings in prediction market history. Post-mortems from the Journal of Prediction Markets and from academic work at Microsoft Research later concluded that the issue was not a lack of information, but a concentration of US-based, college-educated traders whose priors were correlated with the same polling they were using as a signal. The market was not aggregating independent views. It was aggregating a single view, repeatedly.
2022 Midterms: The Red Wave That Wasn’t
By November 2022, the lesson of 2016 was supposed to be learned. It was not. PredictIt and Kalshi (then in its CFTC-approved event contract pilot phase) priced Republicans at 82% to take the Senate and 94% to take the House, with implied seat totals of 54 GOP Senate seats and 240 GOP House seats. The actual result was a 51-49 Democratic Senate and a 222-213 Republican House — a swing of five Senate seats and roughly 18 House seats from the market consensus. The miss cost directional traders an estimated $40 million in realized losses across the two platforms. The structural issue, again, was correlated priors: a flood of late October money from traders anchored to generic ballot polling that systematically understated Democratic enthusiasm in Senate battlegrounds.
2024 Iowa Caucuses: The DeSantis Trap
A smaller but instructive miss came in January 2024. Polymarket priced Ron DeSantis at 24% to finish second in Iowa against Nikki Haley at 68%. DeSantis finished second by 2.1 points. The error was not catastrophic in absolute terms, but it was a reminder that low-information, low-turnout contests with retail-heavy order books tend to overweight the most recent polling cycle and underweight ground-game indicators that do not show up in cross-tabs.
What the October 2026 Board Looks Like Now
Applied to the current board, three contracts stand out as structurally similar to past upset setups. Kalshi’s “GOP holds Senate” market at 54% is pricing a slim but confident majority, with the bulk of positioning concentrated in the final two weeks — a profile that in 2016, 2020, and 2022 preceded sharp late repricings. Polymarket’s “Recession declared by end of 2027” contract at 31% has been drifting lower all month despite three consecutive soft payroll prints, a divergence that mirrors the 2007 and 2019 patterns where the market lagged the underlying data by roughly 90 days. And the “Bitcoin above $150K by December 31, 2026” contract at 42% is pricing a round-number resistance level that historically has acted as a magnet rather than a ceiling in low-volatility regimes.
None of these are predictions that the market is wrong. They are flags that the market’s confidence is being underwritten by the same kinds of correlated priors that produced the misses above. Traders who have made money on prediction markets long-term tend to treat 80%+ favorites in politically or ideologically charged binary contests as positions to fade on size, not to buy.
Where to Trade
The two deepest-liquidity US-accessible venues for the contracts above are Kalshi, which carries the full US political and economic board under CFTC oversight, and Polymarket, which carries the deepest crypto, geopolitical, and 2028 primary contracts. For a side-by-side of fees, limits, and contract coverage, see our best prediction markets rankings, updated weekly.
History does not repeat on prediction markets, but it rhymes. The 88% Brexit price, the 85% Clinton price, and the 82% GOP Senate price all shared a common structural signature: deep liquidity, confident consensus, and a correlated trader base. When you see all three in the same contract, size down.