Best liquidity management in 2026

For six of the eight, no. Gamma did not merely fail to publish the number, it built it and switched it off: the averageReturns feed still ships av_imp_vs_hodl_usd, av_imp_vs_hodl_deposited, av_imp_vs_hodl_token0 and av_imp_vs_hodl_token1 for 91 hypervisors across 7 periods, and all 2,548 of those slots are null, while the dedicated daily and weekly endpoints that once served it now return ["This endpoint is no longer valid."]. Kamino is the one manager publishing enough to check, and checking it is unflattering: its hourly series returns 24,577 records back to 2023-10-09 with share price and both token prices, and reconstructing the comparison for the 90 days to 2026-08-11 gives a vault return of -16.679% against -12.843% for holding the opening inventory — depositors finished 3.836 percentage points behind. Over that same period the same feed advertised totalApy 5.50% and apy30d 5.93%, and its profitAndLoss field read 0. Charm publishes the best versus-hold disclosure in the category and illustrates it with a vault that has been empty for three years: the worked example claims 8.24% "better than holding the assets from 17th February to 25th June" and links to a vault that returned totalSupply() = 0 and lastTimestamp() = 7 July 2023 when read on 2026-08-11.

Not on the two largest published policies. Beefy states that "the range is reassessed and reset every 6 hours" via a privileged moveTicks() call; on a sample of 120 active Base strategies at block 49829563 the median range had not moved in 358.2 hours. 119 of the 120 were older than six hours, 113 older than a day, 25 older than a month, and one had not moved its range since 2026-01-16, 4,959 hours earlier. Median harvest age on the same 120 was 12.3 hours and 119 of 120 had harvested within a day — fees were being collected, and the 9.5% performance fee charged, while the tick boundaries sat still, on a design whose central claim is that impermanent loss stays unrealised only while the position remains in range. Charm documents permissionless rebalancing — "Anyone can call rebalance, unless the vault manager calls" setRebalanceDelegate — and then gates it on most of its live book: 62 of 78 live Ethereum vaults, 79%, and 25 of 59 sampled live Base vaults return a non-zero rebalanceDelegate, which by Charm's own definition leaves only that delegate and the manager able to rebalance. The exception is the rule.

On three of the eight it does not, and the gaps run in both directions. Kodiak publishes a manager fee of 10% and operates 26 funded Islands that take 99% or more: enumerating all 464 Islands from the published factory at Berachain block 24717314 found 165 live, of which 128 carry managerFeeBPS() = 1000, 11 carry 500, 24 carry 9900 and 2 carry 9999 — and Kodiak's own reference defines the arithmetic, "Fee = (Total Fees Earned * managerFeeBPS) / 10000", making those last tiers 99% and 99.99% of LP fees earned. This is not a fringe: 22 of the 24 Islands at 9900 name the factory's own owner and treasury as manager, among them stablecoin pairs of exactly the shape a conservative depositor would choose. Gamma's docs say "Gamma charges between 14-20% of earned fees per vault" while the hypervisor holding $394,795 of the $879,807 on its mainnet feed returned fee() = 2 at block 25731532, which under the contract's own payout arithmetic is 50% of collected swap fees — one byte, changeable by the owner alone, with no timelock and no interface surfacing it. Charm's headline is 2-5% and the chain says a depositor can be paying 21%: 1% protocol fee identically on Ethereum, Base and Arbitrum, stacked under a per-vault manager fee capped at 20%, with 21 of 137 live vaults sampled sitting at exactly that maximum.