Active Liquidity On Chain: Evidence from PropAMMs Across Chains

• PropAMMs emerged in 2024 with operator-set inventory and prices revised through contract updates, addressing the adverse selection faced by passive AMM liquidity providers.Source: §1 · p. 2
• The paper presents a longitudinal measurement study of propAMMs on Solana, Base, and Monad, examining their edge over AMM liquidity providers and its effect on retail flow.Source: §1 · p. 2
• Two seconds after a fill, propAMMs earn 0.37 bps on Solana and 1.19 bps on Base, while AMMs lose 0.22 and 0.62 bps.Source: §1 · p. 2
• The authors identify four sources of propAMM edge: frequent inexpensive updates, counterparty-based pricing, winning cross-venue arbitrage legs, and spoofing through worse execution than quoted prices.Source: §1 · p. 2
• On non-moving-reference-price fills, propAMMs collect 0.26 bps versus AMMs’ 2.59 on Solana, 1.26 versus 1.38 on Base, and 1.62 versus 8.60 on Monad; retail is a larger share of propAMM notional on all three chains.Source: §1 · p. 2

2.1 Automated Market Makers

• AMMs price trades from pool reserves under an invariant, with prices moving after trades; arbitrage aligns pool prices with external markets and exposes providers to adverse-selection costs.Source: §2.1 · p. 3

2.2 Proprietary Automated Market Makers

• In a propAMM, the operator reprices liquidity directly through parameter updates without a swap; the decoded Tessera, Metric, and Elfomo implementations illustrate logic that may differ across undecoded implementations.Source: §2.2 · p. 3
• Tessera places two directional liquidity ladders using an anchor price, spread, and level multipliers, so changing the shared anchor and spread reprices levels together rather than updating each level individually.Source: §2.2 · p. 3
• In the illustrated Base block, Tessera raises its anchor price and widens its spread before a WETH sale; netting the incoming WETH makes earlier ask levels available again while the bid side stays unchanged.Source: Figure 1 · p. 4
• Tessera tracks traded amounts in each direction to derive remaining liquidity, and applies an output penalty that depends partly on quote age, trade source, and transaction priority fee.Source: §2.2 · p. 4

• Studies of Uniswap v2- and v3-style pools find that providers generally lose relative to simply holding deposited assets, while other work studies strategic liquidity provision to reduce losses.Source: §3 · p. 4
• LVR isolates the economic cost of AMM price staleness and arbitrage [30]; later work extends its framework and measures losses that frequently exceed provider fees [32].Source: §3 · p. 5
• Prior designs reduce LVR while keeping liquidity provision passive, whereas propAMMs rely on active liquidity provision and quote updates to avoid adverse selection.Source: §3 · p. 5
• Earlier work measures CEX–DEX arbitrage and its value extraction from AMM providers, while this paper measures markouts from propAMMs’ perspective and finds them positive.Source: §3 · p. 5

• The measurement period runs from 1 September 2025 to 31 August 2026, or begins later for propAMMs launched after the start date.Source: §4 · p. 5
• The researchers collect propAMM trades and updates alongside AMM swaps on Solana, Base, and Monad.Source: §4 · p. 5
• Bybit order book data is also collected for comparison.Source: §4 · p. 5

Solana

• On Solana, SOL/USDC propAMM volume stayed 3.2-to-3.0 times AMM volume, while propAMM volume fell from $1.82 billion to $530 million between the first and last 90 days.Average daily propAMM volume was $1.12 billion, compared with $292 million on AMMs.Source: §5 · pp. 5–6, Figure 2 · p. 6
• On Solana, HumidiFi averaged more than 50% of daily propAMM volume in the first five months; from early 2026, BisonFi led with a 40.3% average share.Tessera averaged 16.5%, SolFi V2 7.4%, and every other propAMM remained below 7%.Source: §5 · p. 6, Figure 2 · p. 6

Base

• On Base, ETH/USDC propAMM share rose from 2.4% to 10.9% as daily volume grew from $9.5 million to $29.6 million; Tessera remained the largest propAMM by total volume.PropAMMs launched on Base on 30 October 2025, two months into the measurement period.Source: Figure 2 · p. 6, §5 · p. 7

Monad

• On Monad, MON/USDC propAMM volume rose from $2.1 million to $3.9 million, while its share of volume versus AMMs increased from 12.5% to 45.7%.Monad launched during the measurement period, and its first propAMM launched in early April.Source: Figure 2 · p. 6, §5 · p. 7

6.1 AMM Markouts

• At 2 s, AMMs lose 0.22, 0.62, and 2.06 bps on Solana, Base, and Monad as their markouts turn negative around fills.The authors relate the markout shape to stale quotes and LVR; differences in magnitude partly reflect token volatility.Source: Figure 3 · p. 8, §6.1 · p. 8

6.2 PropAMM Markouts

• At 2 s, propAMMs earn 0.37, 1.19, and 1.69 bps on Solana, Base, and Monad; AMMs lose 0.22, 0.62, and 2.06 bps.PropAMMs reprice as the reference price moves, while AMMs wait for a trade and can be picked off at stale prices.Source: Figure 4 · p. 8, §6.2 · p. 9

7.1 Frequent and Inexpensive Quote Updates

• Updates cost over an order of magnitude less than swaps on Solana and 5 to 15 times less on EVM chains; Tessera’s update counts correlate 0.56 with ETH realized volatility.Tessera’s Base quote also applies an increasingly worse freshness penalty after a grace period and stops filling once expired.Source: §7.1 · pp. 9–10

7.2 Pricing by Counterparty

• Non-aggregator flow receives wider realized spreads in 68–83% of matched cells (p < 0.001), except at ZeroFi.The spread premium is 3.59 bps at GoonFi and 1.58 bps at SolFi V2, but 0.10 bps at BisonFi and Tessera.Source: Table 1 · p. 10, §7.2 · pp. 10–11
• Tessera’s volume-weighted effective marginal cost is 1.95 bps versus 0.44 bps nominal, with fees accounting for most of the cost.Trades representing 60.4% of volume pay at most 0.01 bps; the default component collects 56.2% of fee revenue.Source: §7.2 · p. 11, Figure 5 · p. 11

7.3 Arbitrage Flow

• PropAMM–DEX arbitrage earns 0.03 bps on Solana, 0.02 bps on Base, and 1.55 bps on Monad; it represents 5%, 13%, and 6% of notional, respectively.Before these arbitrages, reference prices move by about 4.9 bps on Solana, 1.9 bps on Base, and 7.3 bps on Monad; the propAMM reprices while the public AMM does not, leaving a stale pool for arbitrageurs.Source: §7.3 · p. 12
• Tessera’s single fills earn 1.29 bps at 2 s but would lose about 1 bps without penalties; in propAMM–propAMM arbitrage, the other pool falls about 0.5 bps after the fill.The other propAMM is the stale leg, while Tessera is the leg that repriced.Source: Figure 6 · p. 13, §7.3 · p. 13

7.4 Quoting Accuracy

• Only 39% of Tessera trades match the prior-block quote; execution averages 1.08 bps worse, or 0.56 bps worse volume-weighted.Default-fee decreases concentrate in the final tenth of a block and increases in the first tenth; the authors note they lack historical quote data to corroborate a similar reported gap on Solana.Source: Figure 7 · p. 14, §7.4 · p. 14

8 Retail Execution on PropAMMs

• Fills are classified as quiet when the Bybit weighted mid moves less than 1 bps from 5 s before to 1 s after execution.Quiet fills are treated as retail-like because they arrive independently of market movement; their markout is flat across the measured horizon, so its level approximates the price retail pays relative to the reference price.Source: §8 · p. 15

9 Concluding Discussion

• 0.26 vs 2.59 bps on Solana, 1.62 vs 8.60 bps on Monad, and 1.26 vs 1.38 bps on Base are retail half-spreads on propAMMs versus AMMs.The paper concludes that propAMMs give users better prices and that their edge over AMM liquidity providers does not come from retail.Source: §8 · p. 15

8 Retail Execution on PropAMMs

• Quiet fills account for 32% of propAMM notional on Solana, 43% on Base, and 17% on Monad, versus 9%, 16%, and 8% on AMMs.PropAMMs take a larger share of retail flow on all three chains.Source: §8 · p. 15

9 Concluding Discussion

• PropAMMs update prices cheaply and frequently without trades, charge by counterparty, avoid losing to cross-venue arbitrage, and spoof by quoting better prices than trades receive.The study reports a large propAMM share of DEX volume across Solana, Base, and Monad.Source: §9 · pp. 15–16
• Only 39% of Tessera swaps execute at the quoted price, and the average swap receives 1.08 bps less because aggregators route using stale quotes from the previous block.An on-chain router querying each venue inside the swap transaction would remove this gap and route each trade at the price it actually receives.Source: §9 · p. 16

A Data Collection

• The measurement spans 1 September 2025 to 31 August 2026, with Monad data starting at its 24 November 2025 mainnet launch.Source: A Data Collection · p. 18

A.1 Swaps

• Coverage comprises 448,582,796 Solana propAMM trades, 144,106,459 Solana AMM fills, 148,962,685 Base swaps, and 15,659,039 Monad swaps.Solana update counts are estimates: non-swap program invocations are counted as updates, and quote movement cannot be observed.Source: §A.1 · p. 18

A.2 Markout Timestamps

• Solana timestamps cannot distinguish the two to three slots within a one-second interval; Base timestamps approximate early-block trades, while Monad uses consensus-layer timestamps.On Base, trades in the first 26% of transaction positions receive timestamp b −1.8 s and the rest receive b; intermediate Flashblocks retain timing uncertainty.Source: §A.2 · pp. 18–19

A.3 Reference Prices

• Fills are evaluated against Bybit top-of-book references, with USDT-pair prices converted to USDC using the USDCUSDT midpoint.Source: §A.3 · pp. 19–20

B PropAMM MEV Detection Heuristics

• The arbitrage heuristic classifies atomic arbitrage from trader token-balance changes, while route classification depends on which venues the study can decode.Undecodable swaps are treated as separate pools and assigned to “other”; Monad propAMM–propAMM trades against out-of-scope venues are also noted.Source: B PropAMM MEV Detection Heuristics · p. 20

C.1 Marginal Cost and Fee Revenue

• Marginal costs are defined as nominal and effective half-spreads normalized by their respective bid and ask midpoints; fee revenue is attributed across six components.For WETH received, withheld WETH is valued at the swap’s execution price.Source: §C.1 · p. 21

C.2 Swap Classification

• Tessera’s penalty-group analysis measures the volume-weighted share followed by a reference-price move in the trader’s direction over 15 seconds.Groups overlap when a swap incurs multiple penalties; a separate “none” group contains swaps with zero total penalty.Source: §C.2 · p. 22
• At thresholds above 10 bps, address-, age-, and mode-penalized volumes are followed by trader-direction moves approximately 11.54%, 8.64%, and 1.53% of the time, respectively.Source: §C.2 · p. 22, Figure 10 · p. 22

C.3 Quotes vs. Execution

• On 37% of volume, the quote is exact; when default fees differ, traders receive less than quoted on 94% of volume, while 61% of operator-update volume executes better.Preceding trades barely matter: 59% of their volume receives exactly the quoted output, and 99% of residual-group volume receives less.Source: §C.3 · p. 23
• In the example block, raising the default fee from 0 to 538 ppm widened the post-fee spread from 2.78 to 13.54 bps; resetting it narrowed the spread from 13.70 to 2.94 bps.The figure illustrates price and default-fee adjustments and their causes.Source: §C.3 · p. 23, Figure 11 · p. 24
• Tessera charges no penalties on 42.1% of trades, which account for 47.4% of volume.The figure’s mutually exclusive fee-component combinations cover 99.1% of trades; 42.1% pay nothing.Source: Figure 12 · p. 24

D Tessera vs. Elfomo vs. Metric

• Tessera uses two directional pricing ladders, adjusts them through external updates and directional flow accumulators, then applies the combined execution penalty after traversing available liquidity.Source: D Tessera vs. Elfomo vs. Metric · p. 25
• Elfomo reconstructs ladder capacities from vault balances and combines posted updates with WooFi’s Wooracle price, applying level, inventory, quoteage, and caller adjustments during traversal.A further penalty depends on preceding same-side executions in the block and trade count rather than amount-based flow netting.Source: D Tessera vs. Elfomo vs. Metric · p. 25
• Metric combines a Chainlink ETHUSD reference band with separately posted bid and ask quotes, organizes liquidity into bins, and applies penalties during traversal.Metric and Elfomo incorporate pricing adjustments during traversal, whereas Tessera applies its additional execution penalties after traversal.Source: D Tessera vs. Elfomo vs. Metric · p. 25

E Markouts by propAMM

• Tessera reaches 0.56 bps two seconds after a Solana fill, HumidiFi stays near 0.32 bps, and BisonFi reaches 0.06 bps.Tessera’s markout falls to 0.28 bps by 15 seconds; BisonFi’s continues rising past ten seconds.Source: E Markouts by propAMM · p. 26
• On Base, Elfomo remains at 1.6 bps, while Tessera and Metric reach 1.62 bps and 0.26 bps two seconds after a fill.Tessera and Elfomo earn more per fill than any Solana propAMM shown, while Metric earns less than all but BisonFi.Source: E Markouts by propAMM · pp. 26–27
• On Monad, Hanji’s markout declines from 2.14 bps at the fill to 1.65 bps two seconds later, while Metric declines from 1.88 to 1.72 bps.Both are negative five seconds before the fill: Hanji at −3.63 bps and Metric at −1.99 bps.Source: E Markouts by propAMM · p. 27, Figure 16 · p. 27

F Markouts by Route Class

• At 2 s, single fills earn 0.41 bps on Solana, 0.92 bps on Base and 1.44 bps on Monad; propAMM–DEX legs earn 0.03, 0.02 and 1.55 bps.Both route classes contribute to positive pooled markouts.Source: F Markouts by Route Class · p. 27, Figure 18 · p. 28
• Before fills, propAMM–DEX legs follow reference-price moves of 4.9 bps on Solana, 1.9 bps on Base and 7.3 bps on Monad, versus smaller drift for single fills.Source: F Markouts by Route Class · p. 27

G Retail Execution on AMMs

• At 2 s, AMMs earn 2.59 bps on quiet flow but lose 0.49 bps on moving flow on Solana; Base values are 1.38 and -1.00 bps, and Monad values are 8.60 and -2.99 bps.Quiet fills carry 9% of AMM notional on Solana, 16% on Base and 8% on Monad; moving flow dominates pooled markouts.Source: G Retail Execution on AMMs · p. 28, Figure 19 · p. 28