Galaxy Research reported on October 1, 2026 that 69.2% of the 2.9 million retail accounts trading on Polymarket have finished below break-even, producing cumulative losses of $338.9 million on the international prediction market.

The study draws on Polymarket's complete on-chain trading history, spanning the platform's full life since its 2020 launch.

What did Galaxy actually measure?

Galaxy defined retail accounts as those operating at human pace. The filter separates people clicking buttons from machines that place orders faster than any human could react. Automated accounts were carved out of the retail dataset entirely.

The median retail account lost approximately $3. The middle 50% of traders landed between a $36.64 loss and a $0.40 gain, a distribution that concentrates modest damage across the bulk of the user base while leaving room for small, infrequent wins.

The tails of the distribution tell a sharper story. The top 1% of losers absorbed losses up to $4,804, while the top 1% of gainers banked up to $3,381.

What is driving the underperformance?

Polymarket prices outcome shares between $0 and $1 based on the implied probability of a real-world event settling one way or the other. To break even over time, retail traders must overcome both bid-ask spread and, since early 2026, the taker fee charged on orders that immediately match against resting liquidity on the order book.

Specialist behavior appears to amplify the difficulty. Some 44.1% of accounts concentrated mostly on a single topic, spreading bets across multiple markets within one category rather than across categories.

Sports specialists posted the worst win rate at 25.1%. Traders focused on technology and science fared notably better at 41.2%.

Position size tracked outcomes too. Profitable traders carried a median position of $13.96, compared with $10 for those who did not finish ahead.

What happens after a loss?

Galaxy tracked post-trade churn across the dataset. Following a loss, 15.2% of traders stopped trading entirely. After a win, only 6.1% walked away. Post-loss churn ran roughly 2.5 times the post-win rate.

The asymmetry carries operational consequences for the platform itself. A losing trade pushes engagement risk higher than a winning trade pulls engagement forward.

What are the bots doing?

Automated accounts were excluded from Galaxy's retail analysis. Those accounts collectively netted $246.8 million in profits, according to the report.

That exclusion is structurally important. Including bot activity would have shifted the aggregate outcome, though the report does not disclose what share of total platform volume algorithmic accounts represent.

The analysis spans trading from before Polymarket introduced taker fees in early 2026 through more recent activity, capturing different fee environments in a single dataset and reducing the chance that the loss figure is driven by a single regime.

What does this mean for market structure?

For industry observers, the report frames Polymarket as a venue where a majority of click-and-decide participants fund a minority of profitable specialists and a more profitable automated segment. The on-chain record that enables a population-scale study is the same transparency guarantee the platform markets to its users.

The disproportionate churn after losses will draw attention from market-integrity researchers, who have argued that losses of this scale at the retail level warrant closer review of disclosure, fee design and the way implied probability is framed to non-professional users.

Galaxy Research has not disclosed whether further analyses of the same dataset are planned. The combined fee environments captured in the current study span both pre- and post-taker-fee regimes, leaving open how the early 2026 fee introduction reshaped retail profitability for the next cohort of click-and-decide participants.