Who Trades on Polymarket? What Three 2026 Studies Found

Most Polymarket accounts are small. The typical one makes a few dozen trades of a few dollars each and ends close to where it started. A few percent of accounts place most of the orders, and only about 3% show skill that luck does not explain. Three studies published in 2026 measured this from the public trade record, and they agree more than they differ.

Key findings

• The median user placed 46 trades in six weeks, and the average trade was $6.50 (Pew Research Center, 11,989 wallets).
• 58% of those accounts finished within $100 of break-even. 7% made more than $1,000 and 9% lost more than $1,000.
• 69.2% of 2.9 million human-paced accounts were below break-even, and the median one was down about $3 (Galaxy Research).
• Automated accounts were 4.1% of all accounts, placed 80.8% of orders and were up $246.8 million as a group.
• About 3% of accounts show skill that chance does not explain. Only 12% of the top earners are in that group (London Business School and Yale).

How big is a typical Polymarket trade?

The average trade is $6.50, and the average account traded a little over $600 in six weeks.

Pew Research Center followed 11,989 wallets from 7 May to 19 June 2026. It picked them from accounts that had traded on ten high-volume events, then collected everything those wallets did for six weeks.

Trade size changes with the topic. Accounts that traded mostly sports averaged about $9 a trade. Politics accounts averaged about $6, and crypto accounts under $4.

The wallets people call whales trade thousands of dollars at a time and sit far outside this picture. We cover them in what a Polymarket whale is.

How often does a typical user trade?

The median user placed 46 trades and was active on 10 of the 42 days.

Pew's accounts split into four groups by activity (the shares are rounded, so they add up to 101%):

The busiest group traded on 39 of the 42 days. It did not do better for the effort: its typical member lost about $140, a third lost more than $1,000, and 27% made a similar amount.

Most users also stay in one lane. The median account put about 75% of its trades into a single topic, and 24% traded one topic only. Sports accounts placed a median of 69 trades, crypto accounts 59 and politics accounts 13.

Do most Polymarket traders make money?

Most end close to zero. Over six weeks about 44% of accounts were in profit, mostly by small amounts. Over a wallet's whole life the share with $1,000 or more is 2.4%.

These three counts measure different things, so read them side by side and do not average them.

Pew called the typical result break-even: the average account lost less than $2. Galaxy found the same shape in a much larger group. The middle half of its accounts finished between a loss of $36.64 and a gain of $0.40.

Each loss is small because the stakes are small. Together, Galaxy's 2.9 million accounts were down $338.9 million.

Who takes the profit?

Automated accounts are up as a group, and fewer than 3.5% of accounts take more than 30% of all gains.

Galaxy split accounts by pace. Any account that placed more than 50 orders per active day was classed as automated. That group was 4.1% of accounts, placed 80.8% of all orders and was up $246.8 million.

The London Business School and Yale team asked a different question: whose profit is larger than luck would give. They found 3.14% of accounts with skill of that kind. Together with market makers, that is under 3.5% of accounts, and it takes more than 30% of all gains. The lucky winners, 29% of accounts, hold the other 69%. The 67% of accounts classed as unlucky or unskilled carry the whole pool of losses.

CrowdIntel's lifetime count points the same way. The top 0.1% of wallets, about 1,560 of them, hold 71.5% of roughly $1 billion in total profit.

Is a big profit skill or luck?

Usually it is luck. Only 12% of the top earners are in the group that shows skill.

The London Business School and Yale paper covers 1.72 million accounts, 98,906 events and $13.76 billion in volume over two years. For each account it kept the real sequence of markets, times, prices and sizes, and flipped a coin on whether each trade was a buy or a sell. It did this 10,000 times per account. An account whose real profit beat almost all of those coin-toss versions was classed as skilled.

Three results matter to anyone who follows other traders:

• 29% of accounts made money without skill the test could detect. The paper calls them lucky winners.
• About 60% of lucky winners lost money when tested on a separate set of events.
• 44% of accounts classed as skilled on one half of their events were classed as skilled on the other half too. The authors ran the same test on active mutual funds and got about 10%.

So a profit ranking and a skill ranking are different lists. We saw the same thing in our own pool: of 15,371 wallets we collected, 166 passed four skill checks on 30 September 2026, about one in 93. The details are in what we learned from 15,371 wallets.

What do people trade?

Mostly sports, which draws the most trades per account and has the lowest share of specialists in profit.

Pew found that sports, politics and crypto make up nearly all volume on the site. Among accounts that keep to one topic, sports accounts trade most often.

Galaxy looked at accounts that specialise. Sports was the largest group, 47% of specialists, and had the lowest share in profit, 25.1%. Technology and science had the highest, 41.2%, on a smaller sample.

Large wallets lean the same way. In our report for week 41, the 863 wallets that traded, out of 1,049 in our directory, put 54.5% of their volume into sports and 22.2% into esports.

What can these studies not tell you?

• Every count above is a count of wallets. One person can run several, and a team can run one.
• The periods differ. Six weeks and a wallet's whole life give different shares in profit, and both are correct for what they measure.
• Pew's sample started from accounts that traded on ten high-volume events, so quiet accounts are underrepresented.
• Galaxy's line of fifty orders per active day is a judgment call, as the firm says itself. A person trading fast by hand counts as automated, and a slow script counts as human.
• Galaxy's data runs across the period before and after Polymarket introduced taker fees in early 2026.
• Pew and Galaxy both note that Polymarket's separate US platform is not in the data.

What does this mean if you copy traders?

The base rates are against copying whoever is on top. Most accounts hover near zero, a profit of any size is usually luck, and the skilled group is about 3% of accounts.

That is the reason to filter before copying. We check a wallet for a return of 15% or more, at least 30 settled trades, a drawdown of 25% or less and no single trade that made more than half the profit. The methodology page explains each check, and paper trading lets you test a wallet without money in.

Past results do not predict future ones, for skilled wallets either: fewer than half of them kept the label on new events.