The Truth Behind a 7% Probability—Deconstructing Polymarket’s Disaster Prediction Trap Through Behavioral Economics

On Polymarket, the price of the “2026 Hantavirus Pandemic” contract has dropped from a sell-off peak of 35% to 7%. What appears to be a simple market repricing, at a deeper level, reveals a fundamental paradox in how prediction markets operate.
First, what does a 7% probability actually mean? The checklist of conditions that a Hantavirus pandemic would need to satisfy is daunting: the Andes virus must evolve efficient interpersonal transmission, an asymptomatic transmission mechanism, uncontrolled community spread, and sustained spread across multiple global regions—each of these conditions has, at the scientific level, already been demonstrated as “extremely difficult” to achieve.
Under these conditions, the 7% pricing is no longer an epidemiological probability; it is a market consensus that includes an extreme uncertainty premium.
This line of reasoning directly touches the core paradox of prediction markets. Polymarket has long touted “the wisdom of the crowd,” but a heavyweight study jointly conducted by London Business School and Yale University—covering every Polymarket trade from 2023 to 2025—coldly and completely refutes that narrative with hard numbers: among all 1.72 million accounts, only 3% to 3.14% of traders (about 54,000 accounts) are responsible for driving most of the price discovery. The remaining roughly 97% of users only provide liquidity and trading volume, and are, overall, operating at a loss. About 67% of participants are net losers, and 65% of the “biggest winners” profit more from luck than from skill.
For Hantavirus prediction markets, what does this imply? It means that the key driver that pushed the probability down from 35% to 7% is almost certainly that 3% of skilled capital, rather than the 97% retail-consensus. After the WHO’s risk assessment was released, it was this group of professional, skilled players that rebalanced first and decisively, while most retail traders remained largely trapped in panic and hesitation—missing the best timing for price correction.
Another study from Columbia University reveals a structural problem at another layer: between 2024 and 2025, about a quarter of Polymarket trades show suspicious patterns—frequent wash trading, extremely short intervals between trades, and nearly zero position settlement. In other words, even within the 7% data, there may be false trading volume mixed in.
The psychological game in disaster markets is also full of traps. Historical experience shows that disaster-event prediction markets surge sharply during the panic-accelerating phase, but then decay even faster afterward. About 60% of the “lucky winners” turn into losers after switching to different events. This means that, at the 35% peak, the retail traders who blindly chased “YES” are, in the vast majority of cases, already at a loss. As a reverse reference, as of May 11, a related prediction market on Polymarket asking whether “there will be confirmed cases in the US before May 15” still reports “YES” at 27%, with trading volume of only 139,700. With a narrower time window, it is still trading in a state of “low base probability and high premium,” and the market’s irrational side shows up here again.
By this point, a clear conclusion emerges: Polymarket’s pricing evolution (35% → 9.7% → 7%) is first a lagging reflection of the WHO and authoritative scientific research findings, and only second the result of information-based rebalancing decisions by market participants. There is a noticeable time lag between the two, and it is precisely this lag that provides the skilled 3% with a first-mover information advantage—while making 97% of ordinary participants pay the cost.
#Polymarket每日熱點