Swimming in red: Fluid’s ETH-USDC pool springs a leak

This is a segment from the 0xResearch newsletter. To read full editions, subscribe.

Fluid’s flagship ETH-USDC pool has recorded more than $8.5 million in cumulative losses since launch, according to a Dune dashboard by Paradigm researcher Dan Robinson. The issue: a rebalancing mechanism that not only underperformed during volatility, but appears to be systematically unprofitable, even outside of rebalance windows.

The pool, launched in December 2024, was designed to auto-rebalance liquidity once ETH’s price moved beyond ±10%. That design worked well during periods of modest volatility, but as ETH dropped from around $3,800 to $1,560 and then rebounded to $2,400, rebalancing dragged down liquidity provider (LP) capital. CEX-DEX arbitrage opportunities swamped any trading fee income LPs received.

The markout data paints a grim picture. As the chart shows, LP PnL has been consistently negative, not just during rebalances. This contradicts the assumption that fee income from the concentrated liquidity AMM model would offset rebalancing-related losses. Instead, it appears the entire design structure allowed MEV-extractive trades to bleed LP value, even under normal conditions.

Since the May 11 governance post detailing the problem, Fluid co-founder Samyak Jain has been playing defense.

As Sorella Labs CEO, who goes by @0xvanbeethoven, noted, LPs lost millions in rebalancing-driven losses — effectively subsidizing arbitrage with no meaningful upside.

Critics argue the problem isn’t simply rebalance frequency, but the fact that arbitrageurs are able to consistently extract value, while LPs are structurally disadvantaged.

Fluid’s team has responded to the criticism, drawing parallels to the early days of Uniswap v3. Referencing a similar debate around using markouts as a proxy for LP returns, they point to prior commentary from Uniswap researchers, such as Xin Wan, who argued that short-interval markouts can be misleading and don’t fully account for fee accrual or LP lifecycle behavior.

The planned Fluid v2 upgrade — targeted for June or July — will introduce dynamic fees, permissionless pools, and custom LP ranges. It may also work out the LP issues in other ways.

“There are multiple things that can be added including adding them via hooks on v2,” chief operating officer DMH told Blockworks. “I actually like [the] Angstrom solution but have to study it more,” he added.

As proposed interim relief, the Fluid team suggests distributing 500,000 FLUID tokens (0.5% of supply) over a year to affected ETH-USDC LPs, plus $400,000/month in rewards until DEX v2 launches.

Rather than widening the rebalance band to reduce loss frequency, Fluid also proposes narrowing the range from ±10% to ±7.5%, increasing fee accrual per unit of liquidity, although that would also increase rebalancing events and potential losses.

This short-term fix has raised eyebrows. “If your pool is losing money on average, increasing concentration will increase both losses and variance,” wrote Dan Robinson. “So concentrating liquidity more would be a double whammy.”

Even if DEX v2 gives LPs more control, it assumes users will be able to model volatility risk and pick defensible ranges, a challenge even protocol governance struggled to get right in v1.

It’s to Fluid’s credit that the team has acknowledged the issue publicly, citing their own $1 million LP position as evidence of skin in the game. They also emphasize that correlated pairs (like cbBTC–WBTC) have performed well, and so the ETH-USDC pool’s struggles are not representative of the broader DEX performance.

Still, the damage to LP trust is real. Whether DEX v2 can reverse that dynamic will be the true test of Fluid’s unified architecture.

For now, the ETH-USDC pool stands as a cautionary tale: capital efficiency can cut both ways, especially when market volatility meets rigid strategy design.

Source