Fee APR ranks liquidity pools backwards: realized loss-versus- rebalancing and passive returns on Uniswap v3

Imagine a vast, digital marketplace where people trade digital assets without a central bank or a traditional exchange. In this world, ordinary people can act as market makers by depositing their money into shared pools, earning a small fee every time someone swaps one asset for another. For years, the standard way to find the most profitable pools has been to look at a single number: the annual percentage return generated by fees. It is a simple metric, easy to read on a screen, and it suggests that the more trading volume a pool has relative to the money sitting in it, the more money the provider will make. It feels like a reliable compass for anyone trying to grow their wealth in this new financial landscape.

However, a new study reveals that this compass is pointing in the exact opposite direction of where the money actually is. The research, conducted by Maksym Kovalenko, investigates the real-world earnings of these passive providers across three major digital networks. The study finds that the popular metric used to rank these pools is not just noisy or imperfect; it is actively misleading. When investors follow the standard advice to pick pools with the highest fee returns, they end up losing significantly more money than if they had simply held a random mix of pools or avoided the selection process entirely. The paper demonstrates that the pools appearing most attractive on paper are actually the ones where the hidden costs of trading are highest, effectively draining the provider's capital faster than the fees can replenish it.

To understand why this happens, one must look at the mechanics of how these pools function. When a provider deposits money, they are essentially betting that the price of the assets will stay relatively stable. But in a volatile market, professional traders and automated bots constantly adjust their positions to match the changing prices on other exchanges. Every time the price moves, the pool's inventory becomes misaligned with the true market value. The difference between what the pool holds and what it would be worth if it had been perfectly adjusted is a hidden loss, known in the industry as "loss-versus-rebalancing." This loss is the price the passive provider pays for the convenience of letting others trade against them. The study shows that the standard fee metric fails to account for this cost. In fact, the pools that generate the most fees are often the ones where the price is moving the most aggressively against the provider, causing the hidden losses to overwhelm the earnings.

The researcher tested this theory by tracking 62 different trading pools over a period of 37 weeks in 2026. Instead of relying on the standard fee numbers, the study built a more accurate picture by comparing every single trade in the pool against a precise, second-by-second price from a major centralized exchange. This allowed the team to calculate exactly how much money was earned in fees and exactly how much was lost to market movements. The results were stark. Across the three networks studied, the passive providers earned roughly 120 million dollars in fees but lost over 153 million dollars to these hidden market adjustments. The net result was a loss of nearly 36 million dollars. The study found that on the main network, nearly 76 percent of the weeks were unprofitable for the average provider, and the single most popular trading pair lost money in every single week it was observed.

The most surprising discovery came when the researcher tested what would happen if an investor tried to use the standard fee metric to pick the best pools. The study simulated a strategy where an investor sorted all the pools by their headline fee returns and picked the top group to invest in for one week. This strategy resulted in an annualized loss of nearly 9.4 percent. In contrast, simply holding a random mix of all the pools resulted in a smaller loss of about 5.7 percent. The metric did not just fail to find winners; it actively selected the worst possible options. The pools that looked the most profitable were actually the ones where the hidden losses were the most severe. This pattern held true across all three networks tested, confirming that the problem was not a fluke of a single market but a fundamental flaw in how the metric measures success.

The study also explored whether other factors, such as the volatility of the asset prices or the specific structure of the liquidity, could explain why the fee metric was so misleading. The researcher found that the problem was not simply that volatile pools were riskier. Even when accounting for how much the prices were swinging, the fee metric still ranked the pools in reverse order of their actual profitability. The study also investigated a theory that the physical placement of money within the pool—specifically, how close the liquidity was to the current price—might offer a clue to future earnings. While the researchers mapped out exactly where the money was sitting and how it moved, they found that this spatial information did not help predict which pools would be profitable. The simple fact of how much money was lost to market adjustments remained the strongest predictor of future performance.

Another aspect of the study looked at a phenomenon known as "just-in-time" liquidity, where sophisticated traders briefly inject money into a pool to capture a fee and then immediately withdraw it, often at the expense of the passive providers. The study measured this activity with high precision and found that while it existed, it was not the primary cause of the losses. In 2026, this activity captured less than one-tenth of one percent of the total fees. The real drain on capital was the constant, unavoidable cost of the market moving against the passive provider, a cost that the standard fee numbers completely ignored.

The implications of these findings are significant for anyone participating in these digital markets. The study concludes that the standard way of evaluating these investment opportunities is broken. The metric that everyone uses to find the best returns is actually a signal for the worst returns. The research suggests that the only reliable way to predict future performance is to look at the past performance of the net earnings after accounting for all losses, rather than just the fees. By using this corrected view, an investor could have reduced their losses to nearly zero, whereas following the standard advice led to substantial financial damage. The study does not claim that these markets are broken beyond repair, but it does show that the tools used to navigate them are currently pointing in the wrong direction. For the curious observer, the lesson is clear: in this complex financial ecosystem, the most obvious sign of success is often the most reliable sign of failure.