Community Trust ScoreVerified
A study posted on Sept. 29 found that realized round-trip spreads for SOL/USDC trades were wider for non-aggregator flow at five of six major proprietary liquidity venues on Solana. The research, authored by Ozan Solmaz, Lioba Heimbach, and Jason Milionis, indicates that access to the same liquidity pool did not necessarily result in comparable trading terms depending on how the order was routed.
Methodology and Data Scope
The preprint examined proprietary automated market makers, or propAMMs, on the Solana, Base, and Monad blockchains. Observations for the study ran from Sept. 1, 2025, through Aug. 31, 2026, with later start dates for newer deployments. The core analysis focused on the six largest measured Solana propAMMs for SOL/USDC pairs, comparing flow routed through aggregators against non-aggregator flow.
To estimate realized round-trip spreads, the researchers paired consecutive fills on opposite sides of the same pool. These paired fills had to occur within 25 slots and have trade sizes within a factor of two of each other. The methodology required that no intervening quote update from the operator occur between the paired fills, and both legs of the trade had to belong to the same routing class. Comparisons were then made within cells defined by pool, day, and trade-size bucket.
This approach allows for a more controlled comparison than analyzing unrelated venues or differently sized orders. However, the study notes that the two fills in a pair do not necessarily represent a single customer’s actual buy-and-sell journey. The resulting data represents a historical association between routing class and realized spreads rather than a measurement of what every interface delivers at any given moment.
Spread Differences and Fee Mechanics
The study found that for the first five venues, non-aggregator spreads were wider in 68% to 83% of matched cells. Reported exact sign-test p-values for these venues were below 0.001. ZeroFi was the exception, showing a small negative premium for the matched median, meaning non-aggregator spreads were not wider at that venue.
Specific median figures illustrate the distinction between overall performance and matched comparisons. Tessera’s overall median spread was 0.9 basis points for aggregator flow and 0.93 basis points for non-aggregator flow. However, its median matched premium was only 0.1 basis points. One basis point is defined as 0.01 percentage point. The matched premium is calculated by comparing the two classes inside each cell before summarizing those differences, which differs from subtracting the displayed overall medians.
The economic rationale for these differences often involves stale prices. PropAMM operators can send price updates without executing a swap, reducing exposure to stale-price losses when updates arrive before incoming trades. The study decoded a Tessera implementation on the Base blockchain, identifying six fee components: base, age, priority, mode, default, and address. In this implementation, the transaction sender and the contract calling the interface could affect default and address fee treatment, including exemptions and penalties.
The researchers documented that Base Tessera fees decreased near block ends and increased near block starts. This timing helped explain differences between a previous block’s quote and the eventual fill. The study noted that the researchers lacked historical Solana quote data to corroborate an analogous quote-to-fill gap on that specific chain. Separately, 0x’s March investigation reported Base quote-to-fill discrepancies involving Flashblock timing and spread changes.
Wallet Fees and Interface Costs
Wallet fees and quote-to-fill slippage measure different costs than the historical spread gaps identified in the study. Phantom’s swap disclosure, checked on Nov. 2, lists a 0.85% fee for most swaps, with exceptions for specified Cash-account flows and deposits to a perps balance. These wallet charges are distinct from the round-trip spread measurements in the paper.
Jupiter, a routing provider on Solana, documents a 2-basis-point platform fee for SOL-to-stablecoin swaps on its managed order path. This path allows Metis, JupiterZ, Dflow, and OKX to compete for liquidity. Jupiter’s customizable build path uses Metis only and documents no Jupiter swap fee, though integrator charges and network costs apply. JupiterZ supplies request-for-quote liquidity, which is a separate mechanism from on-chain propAMM pools.
Jupiter describes using a JIT Swap mechanism that re-quotes candidate pools on-chain, allowing the selected pools or split to change if another candidate allocation offers a better output. The study concludes that adding a historical matched spread premium to a current wallet fee would produce a misleading savings estimate, as it would mix a paired round-trip statistic with a charge on a particular swap without establishing which costs were incremental.
Frequently Asked Questions
What did the study find about Solana propAMM spreads?
The study found that non-aggregator trades had wider realized round-trip spreads than aggregator trades in five of six major Solana propAMMs. The p-values for these differences were below 0.001.
How did the researchers define a matched trade pair?
Researchers paired consecutive fills on opposite sides of the same pool if they occurred within 25 slots, had trade sizes within a factor of two, and had no intervening quote update from the operator.
