Who actually wins on Polymarket?

By the Cent Signals editorial desk. Reviewed June 17, 2026.

A 2026 academic study reconstructed the complete public transaction history of Polymarket and asked a plain question: who ends up ahead, and why. The answer is that profits are concentrated in a small group of users, that most participants finish with a loss, and that the behavior most associated with positive performance is providing liquidity rather than trading large amounts. This guide summarizes the findings of Akey, Gregoire, Harvie, and Martineau, "Who Wins and Who Loses in Prediction Markets? Evidence from Polymarket" (2026), available on SSRN. Like everything on this site it is description of public data and published research, not direction.

What the study looked at

The authors assembled the complete public transaction history of Polymarket, about 67 billion dollars of trade volume across 2.4 million users, spanning November 2022 to March 2026. Because the platform settles on-chain, every trade, every price, and every outcome is observable in public records, which let the study measure realized profit and loss for each wallet directly rather than relying on surveys or samples. That completeness is what makes the findings unusual: they describe an entire market rather than a slice of it.

Gains are concentrated and most users lose

The distribution of profit was steep. Among users with positive profit, the top 0.1% captured 51.2% of all gains, and the top 1% captured 76.5%. On the other side of the ledger, roughly 69% of all users ended the sample with a loss. These figures fit together because each contract is zero-sum: before fees, every dollar of gain on one side is a dollar of loss on the other, so a heavy concentration of winnings necessarily implies a broad base of losses.

Liquidity provision predicts profit better than size

The strongest single predictor of positive performance was not how much a wallet traded but how it traded. The study found that a wallet's maker volume share, the fraction of its activity posting limit orders that add liquidity, was the most powerful behavioral signal. A one-standard-deviation increase in maker share was associated with a 9.0 percentage point higher probability of positive performance, the largest behavioral effect the study estimated.

The contrast across the distribution is stark. The top 0.1% of earners supplied 47.3% of their volume as makers, versus 17.1% for the bottom 95%. The implication the data points to is that transacted volume by itself is not skill; the act of posting resting orders and absorbing the flow of others was far more closely associated with finishing ahead. For what the activity figures themselves measure, see volume vs liquidity on Polymarket.

Prices are accurate, skill is real but limited

On aggregate, Polymarket prices were well calibrated: a contract priced at p resolved in its favor about p percent of the time. That is the market working as intended, and it sits alongside the favorite longshot caveat described in the favorite-longshot bias. At the level of individual traders, forecasting accuracy varied widely. Per-trade excess hit rate ranged from about minus 15 to minus 18 percentage points in the loss tail to about plus 20 points for the most profitable users.

Skill, in other words, is real but not widespread. A finite-mixture decomposition classified only about 29% of traders as skilled, which means the bulk of activity came from participants whose results were consistent with noise rather than durable edge.

Performance fades and the crowd churns

Month-to-month performance was only modestly persistent, and what persistence existed concentrated among limit-order traders. The population also turned over quickly: about 44% of users stopped trading within a month, rising to as much as 66% within six months, a group that included 55% of the best performers. The study notes that apparent persistence may therefore reflect selection, who keeps showing up, rather than durable skill.

The authors also looked closely at the 100 most successful accounts. They found those gains consistent with forecasting skill and skilled liquidity provision, and not with insider trading. The picture that emerges is of a market where a small share of accurate forecasters and active liquidity providers earn most of the profit, while a large and shifting crowd of short-lived participants supplies it.

How Cent Signals reflects this

Because the study found realized profit to be the cleanest measure of who came out ahead, Cent Signals now ranks wallets by realized profit drawn from the same on-chain records the paper used. You can read the traders by realized profit ranking, and the methodology page describes how those figures are computed. These rankings are a description of what the public ledger already shows, nothing more.

Frequently asked questions

Who wins on Polymarket?

The study by Akey, Gregoire, Harvie, and Martineau found that gains were highly concentrated: the top 0.1% of users with positive profit captured 51.2% of all gains, and the top 1% captured 76.5%. The wallets that did best were associated with sustained forecasting accuracy and, more strongly, with supplying liquidity by posting limit orders rather than simply transacting large volume.

Do most people lose money on Polymarket?

According to the same study, roughly 69% of all users ended the sample period with a loss. Because each contract is zero-sum, one side's gain is the other side's loss before fees, so the data shows that aggregate profit accrued to a narrow group while the majority of wallets finished negative.

Does trading more make you more profitable on Polymarket?

The data does not support that reading. Transacted volume by itself was not a reliable predictor of positive performance. The strongest single behavioral predictor the study estimated was a wallet's maker volume share, the fraction of activity that posted limit orders adding liquidity. A one-standard-deviation increase in maker share was associated with a 9.0 percentage point higher probability of positive performance.

Is the study evidence of insider trading?

No. The authors examined the 100 most successful accounts and found their gains consistent with forecasting skill and skilled liquidity provision rather than insider information. The paper frames the concentration of profit as a feature of skill, liquidity provision, and selection, not as evidence of insider trading.