A preprint released on September 29 examined realized round‑trip spreads for SOL/USDC trades on Solana’s six largest proprietary automated market makers (propAMMs). The analysis covered data from September 1, 2025, to August 31, 2026, and compared trades routed through aggregators such as Jupiter with those sent directly to the propAMMs.
Methodology
The researchers paired consecutive fills on opposite sides of the same pool, limiting the time between fills to 25 slots and ensuring trade sizes were within a factor of two. Both legs of each pair had to belong to the same routing class (aggregator or non‑aggregator) and could not be preceded by a quote update from the pool operator. Comparisons were made within cells defined by pool, day and trade‑size bucket.
Key Findings
- For five of the six venues, non‑aggregator trades showed wider median spreads than aggregator trades in 68 % to 83 % of matched cells.
- Median spreads for aggregators ranged from 0.09 to 0.9 basis points, while non‑aggregator spreads ranged from 0.71 to 4.9 basis points.
- The matched premium – the median difference between the two routing classes within each cell – was positive for all venues except ZeroFi, which showed a small negative premium.
- Statistical tests reported p‑values below 0.001 for the five venues with wider non‑aggregator spreads, indicating strong significance.
Interpretation
The results suggest that accessing the same on‑chain pool does not guarantee comparable trading terms; the route a trade takes influences the price received. Proprietary venues can update quotes without executing swaps, potentially reducing stale‑price losses for operators. Aggregator routing, which searches across multiple venues and can split orders, was associated with tighter spreads in most cases.
Additional Context
Wallet fees and platform charges were noted as separate from the measured spreads. For example, Phantom disclosed a 0.85 % fee for most swaps, while Jupiter listed a 2‑basis‑point platform fee for managed SOL‑to‑stablecoin swaps. These fees affect the total cost to traders but are not reflected in the round‑trip spread metric used in the study.
The authors also highlighted fee structures on Base’s Tessera implementation, where multiple fee components (base, age, priority, mode, default, address) can vary based on the transaction sender and contract caller. This demonstrates how proprietary venues may incorporate caller identity into pricing.
Conclusion
The study underscores the importance of routing decisions in determining trade costs on Solana. While aggregators generally delivered tighter spreads, traders should consider both routing paths and wallet/platform fees when evaluating overall transaction costs.
