It's almost a cliché now: prediction markets called Brexit when the polls didn't. They called the 2016 US election uncertainty when the pundits didn't. They've outperformed FiveThirtyEight on plenty of races.
But there are also nights when the polls were closer. Pretending otherwise is selling you a story, not analysing the data.
Why Markets Usually Beat Polls
- →Polls capture stated intent. Markets capture revealed conviction (with money on it).
- →Polls have known biases (response rates, weighting). Markets aggregate across them.
- →Polls update slowly. Markets update by the second.
- →Polls don't price in news. Markets do, automatically.
- →Polls are noisy on the tails. Markets compress noise via aggregation.
Hanson, Wolfers, and several other forecasting researchers have shown markets typically have lower forecast errors than poll aggregates over multi-cycle datasets.
When Polls Beat Markets
Quietly, the cases are real:
- →Very low-liquidity markets where prices are essentially one trader's opinion
- →Events where the relevant polls have huge sample sizes and the markets have small ones
- →Surprise events that the market hasn't had time to absorb (the first 24 hours after major news)
- →Markets dominated by partisan or motivated traders pushing prices for non-information reasons
- →Local races below the threshold where serious money shows up
The Real Question: How Do You Use Them Together?
Pretending markets and polls are rivals is a journalist's frame, not an analyst's. They're complements.
Polls are the input. Markets are the integration. Polls measure one slice of opinion at one time. Markets weigh polls plus everything else: news, gossip, momentum, the bookmaker odds, the size of the bets coming in.
Use polls to understand why a market is priced where it is. Use markets to understand what the world is currently betting will happen.
How to Read a Disagreement
When the polls say 50/50 and the market says 65/35, one of three things is true. Either the market knows something the polls don't. Or the market is wrong and there's an opportunity. Or polling has a known bias the market is correcting for.
Your job as a trader is to figure out which one. If you can't, the market is probably right.
The 2024 Stress Test
Across the 2024 UK general election, the Premier League title race, and the US presidential race, prediction markets were closer to the final result than the polling average in every case — but only modestly. The margin of victory was inches, not miles.
That's the honest takeaway. Markets are a sharper instrument than polls. They're not magic. The wisdom of the crowd is real but it's not infinite.
"Markets aren't always right. They're more right, more often, with less excuse-making when they're wrong."
— Boromarket