Polymarket smart money only 3.14% of users, study finds
A new academic study of Polymarket trading data from 2023–2025 argues that Polymarket’s “smart money” is a small minority.
Researchers from London Business School and Yale analyzed about $13.76B in volume across 210,322 markets and 1.72M accounts. They classify only 3.14% of accounts as “skilled winners.” Despite being under 3.5% of users, this group (including market makers) captures over 30% of total gains.
The paper also challenges whether profits reflect true skill. In robustness tests that randomly reversed historical buy/sell directions 10,000 times, only 12% of top earners overlapped with the skilled group. Around 60% of “lucky winners” later flipped to losers in an out-of-sample check, while “unskilled/unlucky” accounts absorbed about 67% of losses.
On information speed, the study finds only skilled traders adjust order imbalance in line with surprises around major announcements (e.g., FOMC and earnings).
Later analysis flags potential insider-like behavior: 1,950 accounts traded around specific events and then went inactive, moving prices 7–12x more per dollar than skilled traders. However, the authors say this appears too event-specific to fully explain Polymarket’s overall price discovery.
For traders, the key takeaway is that Polymarket pricing may rely more on a concentrated set of informed participants than broad participation—affecting liquidity expectations and signal reliability, especially during high-attention events.
Neutral
The study’s main impact is on how traders should interpret Polymarket price discovery rather than on any direct token valuation. By showing that only 3.14% of accounts drive most gains and that profits may include “luck,” it suggests signals can be concentrated and potentially less robust for broad participation.
In the short term, this could lead traders to tighten filters (e.g., watching order flow from the identified skilled cohort, especially around FOMC/earnings) and adjust expectations for how quickly prices reprice information. That can improve execution quality but may also increase skepticism about trades based purely on winners.
In the long term, if market participants increasingly react to this evidence, liquidity composition and trading strategies around high-attention events may change, but the paper does not provide evidence that overall price discovery is systematically manipulated. The flagged insider-like accounts appear too event-specific to overturn the broader finding.
Net effect on the crypto market in question is therefore neutral: it reshapes trading strategy and signal confidence for Polymarket-based pricing, without clear directional implications for market-wide prices.