Study: ~3% of Polymarket accounts drove most price discovery from 2023 to 2025, suggesting market accuracy comes from the informed minority, not crowd wisdom

CoinDesk 2026-04-28 Sam Reynolds

Context & Ripple Effects

The finding arrives amid related analyses showing that Polymarket’s returns are even more concentrated than participation: one study of 1.6 million accounts found that 0.1% captured 67% of profits. Other coverage has raised questions about wash trading and suspiciously profitable activity, making it important to distinguish informative trading from visible volume.

Together, this shifts the relevant question for prediction-market platforms from whether a large crowd participates to whether a small set of informed traders can trade on fair, credible market infrastructure.

First-order effects

Polymarket’s apparent price-discovery performance is attributed primarily to a small minority of accounts, rather than to broad user participation.
The study gives users and observers a reason to weigh trader quality and market integrity more heavily than headline activity when assessing Polymarket’s signals.

Second-order effects

Platforms and market-data users face stronger pressure to identify whether prices are being set by informed trading, concentrated repeat traders, or potentially manipulative activity.
The combination of concentrated profits, wash-trading allegations, and suspicious-profit findings makes credible surveillance and transparent market-quality metrics more consequential for prediction-market adoption.

Third-order effects

If price discovery remains concentrated, prediction markets may function less as pure aggregators of crowd wisdom and more as distribution channels for the information held by a specialized trading class.
That model can still produce useful forecasts, but its legitimacy will increasingly depend on whether platforms can show that concentrated influence reflects skill rather than unfair informational or market-structure advantages.

The trend: Prediction-market platformization is moving toward a market-quality test: broad access matters, but durable trust depends on who sets prices and how credibly those trades are policed.

Discussion

@bonkdacarnivore @bonkdacarnivore on x
I've been on this for a while. “Prediction markets” are just a way for insiders to trade non-public information, which is a problem all its own. The larger issue, however, is that in a society where we've either gamified everything, turned it into a casino or made everything a
@peterwildeford Peter Wildeford on x
People like me have been behind correctly calibrating all future events. It's been thankless work that's been wrongly attributed to the “crowd” (who are idiots). I'm happy to finally be recognized. You're welcome.
@robinhanson Robin Hanson on x
Yes that's the way it has long worked in most financial markets. We could instead directly subsidize such markets, in which case we wouldn't have to tax all those uniformed traders.
@benshindel Ben on x
(1) yes, this is how markets work (2) rediscovers the pareto principle: obv a small % of people will capture a large % of value (3) how is this “not the wisdom of crowds”? 3% still = 1000s of traders (4) with *only* the informed traders, the market would be far less accurate!
@barneyflames @barneyflames on x
More proof that prediction markets are fake, the information content mostly comes from insider trading, not wisdom of the crowds.
@tunguz Bojan Tunguz on x
Interesting. Still bullish on betting markets as a prognostication tool. But they do take a heavy toll on average an average user.
@rgomezcram Roberto Gomez Cram on x
Polymarket prices are highly accurate in predicting future events. The source of that accuracy is less obvious. In a new working paper, we find it is not the “wisdom of crowds,” but a small minority of informed traders. Fewer than 3% of accounts appear to drive price discovery; […

Chronicles