Pump.fun: Why Displayed Gains Can Mislead Copy Traders

A Bitquery study reveals a significant gap between the gains displayed on Pump.fun and the profits actually derived from trading. Out of 752.5 million dollars analyzed, only 193.9 million come from purchases followed by sales. The rest is mainly explained by token transfers and suspicious operations.

Pump.fun, displayed gains that do not reflect actual profits

The platform continues to generate revenue, with $18.6 million despite the memecoin crash. But the figures displayed by traders do not always tell the whole story. In its investigation published on October 22, Bitquery analyzed 3,170 positions larger than $100,000, closed between October 30, 2025, and August 30, 2026.

The observation is striking: only $193.9 million, or 26% of the total, come from wallets that bought tokens before reselling them. However, 51% of the displayed gains come from selling tokens received from other addresses. 

Moreover, about 13% correspond to wash trading loops, where linked wallets exchange tokens among themselves. The remaining 10% do not fall into any of these categories.

Token transfers that distort displayed gains

This difference is partly explained by how profits are calculated. Tracking tools generally subtract recorded purchases from sales. Thus, when a wallet receives tokens for free and then resells them, its history can show a gain with no acquisition cost. Yet someone may have paid for these tokens upstream.

Bitquery identified 1,579 positions of this type. In 1,104 cases, transfers trace back to the token creator, either directly or via other wallets.

The UOTF example illustrates this mechanism: four temporary addresses each received nearly eight million tokens before transferring them to selling wallets as part of a coordinated operation. One of these addresses subsequently sold $1.59 million worth of tokens.

This mechanism complicates copy-trading. A tool intended to replicate a trader’s purchases cannot follow a position if it only sees the sale without knowing the initial entry price.

Copy-trading: fees and data that complicate replication

Some platforms try to flag these situations. The GMGN documentation specifically mentions “Didn’t buy” and “Sold > Bought” alerts, indicating a wallet received tokens without purchasing them or sold more than it had acquired.

However, these alerts do not guarantee that another user can replicate the displayed returns. GMGN’s transaction fees, set at 1% excluding network and priority costs, along with slippage and execution delays, further reduce potential gains.

Finally, the study has two limitations: Bitquery did not calculate the actual returns of copy-traders and excluded 2,943 six-figure positions closed after August 30. The share of gains truly accessible to followers remains to be determined.