Prediction Markets Are Betting on the Wisdom of Crowds. The Crowd Hasn't Fully Shown Up Yet.
Prediction markets can work brilliantly when attention converges. The problem is that, for most contracts, too few people are there to make the crowd wise.
There's a seductive idea behind prediction markets: put real money on the line, let enough people trade against each other, and the resulting price will beat pundits, polls, and even experts. It's the "wisdom of crowds" thesis, and it's the whole reason platforms like Polymarket and Kalshi get taken seriously by traders, journalists, and increasingly Wall Street itself.
That thesis isn't wrong so much as unfinished. A recent CNBC analysis is a useful gut check on how much of the market has actually caught up to the theory and right now, the honest answer is: not all of it, not yet.
Where the theory meets the data
CNBC dug into Polymarket's own data going back to 2021 and found that roughly seven out of every ten closed markets on the platform saw less than $10,000 in total reported volume. Fewer than one in ten markets ever crossed into six-figure territory. Nearly 5% of all closed markets, more than 45,000 of them, recorded no trading volume whatsoever. Kalshi, Polymarket's chief rival, shows a similar pattern in a parallel analysis of its on-chain activity.
None of that means prediction markets don't work. It means the wisdom-of-crowds mechanism is unevenly distributed and concentrated in a minority of contracts that draw real attention, while a long tail of markets exists more in theory than in practice.
Why thin markets ask harder questions of the theory
The wisdom-of-crowds effect depends on scale and diversity of participation. A handful of trades doesn't average out noise as effectively; thinner participation gives more weight to whoever happens to show up. Academics quoted in the CNBC piece make a related point: thin markets tend to be more volatile, since a single trade can swing a price that's supposed to represent a stable probability. Spreads widen too, which quietly taxes anyone trading in a shallow contract, particularly newcomers who don't know to check volume before clicking buy.
Then there's the bot question. One researcher's on-chain analysis found that in markets under $10,000 in volume, bots accounted for the large majority of trading activity. These aren't degenerate gamblers moving prices for fun, they're automated strategies extracting small, consistent profits, and by the researcher's estimate they pulled in over a million dollars from shallow markets alone versus tens of millions from the deep ones. Retail traders, meanwhile, were shown to lose money in both thin and heavy markets. It's worth noting bots don't appear to be pushing prices away from fair value on purpose – their incentive is profit, not manipulation – but their outsized presence in low-volume contracts is still a sign that the "crowd" setting those prices is thinner and less human than the wisdom-of-crowds story assumes.
The case that volume isn't the whole story
It's worth taking seriously the researchers who argue volume alone is the wrong lens – that what matters is who is trading, not how much. Research cited in the CNBC piece, including work out of Yale and the London Business School, found that informed or skilled traders were responsible for most of the accuracy gains on Polymarket, somewhat independent of a market's overall size. A statistics professor at Rutgers adds that low liquidity, on its own, doesn't discredit a market's signal, it just means that signal should be weighted with the context in mind.
That's a genuinely important qualifier, and it complicates any simple "thin equals unreliable" narrative. But it also comes with a practical catch: a retail trader scrolling a list of contracts has no easy way to tell whether a given market's thin volume is being shaped by informed, skilled traders or by a handful of bots and a rounding error. Nothing on the interface distinguishes the two. So even if the theory holds up in aggregate, it's hard for an individual user to lean on it with confidence in any single market.
Meanwhile, the markets that clearly do function as advertised – the ones the analysis found clustered around short-dated, high-attention events tied to elections, geopolitics, or major public figures – show the concept working exactly as intended: high volume, short time horizons, broad participation, real price discovery. That's the wisdom of crowds in action. It's just not yet the median experience across the platforms.