A new Galaxy Research analysis of Polymarket's on-chain records found that more than 69% of retail trading accounts finished with a loss. Across the 2.9 million accounts studied, combined losses came to $338.9 million.
How the study was done
Galaxy used Polymarket's full on-chain history, looking at positions, entry prices, holding periods and payouts. It removed 125,429 accounts that averaged more than 50 orders per active trading day, treating them as likely automated. Those wallets made up 4.1% of the total but placed 80.8% of all orders.
| Account type | Accounts | Share of wallets | Share of orders |
|---|---|---|---|
| Likely automated | 125,429 | 4.1% | 80.8% |
| Human-paced (analyzed) | 2.9 million | 95.9% | 19.2% |
What the data showed
- The median retail account lost about $3, though a minority lost thousands.
- After a loss, 15.2% of accounts did not trade again within 30 days, versus 6.1% after a win.
- Both winners and losers tended to return with slightly smaller positions, but winners cut risk less than losers did.
- About 44% of traders specialized, putting over 60% of activity into one topic. Of these, 28% finished profitable, compared with 30.4% of generalists.
- Sports specialists were the largest group and had the lowest profitability rate. Tech and science specialists did better, with 41.2% profitable. Galaxy suggested deeper subject knowledge may play a part, but the data does not prove cause.
- Profitable traders generally placed bigger bets and traded more often. Holding time showed no clear link to profit.
Limitations
The study covered Polymarket's international platform, not its separate US exchange. It also tracked wallet addresses rather than people, so one trader using several wallets could be counted multiple times.
Legal backdrop
Prediction markets face growing legal pressure. In the US, some officials argue that contracts on sports and other uncertain outcomes resemble gambling. Baltimore sued Polymarket and Kalshi in August, alleging unlicensed sports betting. New York sued Polymarket's US arm in September, alleging unlicensed gambling and allowing users aged 18 to 20 to trade despite a 21-year minimum for mobile sports betting. In South Korea, authorities opened cases against 26 users and referred 18 to prosecutors over about $12.7 million in bets.
Why it matters
For anyone learning about crypto, the findings show that prediction markets are not a reliable way to make money for most ordinary participants. Losses were common, and losing appeared to make people more likely to leave. The numbers also show how much activity can come from automated accounts. Finally, the legal disputes are a reminder that the rules for these platforms are still unsettled and differ by location.
Originally reported by 



