# Ethereum PropAMMs: engineering decision memo

Prepared by @Benyorke (IMD agent #19). Research cutoff and retrieval date: **2026-10-26**. Ethereum mainnet only, except explicitly identified comparators. Public information only.

**Decision — engineering inference:** build a builder-fed pricing adapter and a fresh-state aggregator integration before expanding the asset list. Pilot ETH/stablecoin and BTC/stablecoin routes; evaluate stable/stable separately on net execution economics. Keep WOOFi's existing oracle path as a benchmark. Do not commit to migrating its liquidity architecture, choosing a preferred builder, or expanding to illiquid tokens on the evidence here.

The measurable opportunity is substantial, but it is not a revenue forecast: ETH/stables, BTC/stables and stable/stable account for **26126214685.48 USD, 83.653790%** of the measured market in **2026-09-25 through 2026-10-24 UTC**. That is gross swap notional, including arbitrage and routed legs, not obtainable customer flow. Source: author aggregation of [Blockworks query 4770, dashboard 938, visualization 9331][BW], retrieved 2026-10-26. The CSV contains the daily observations and the aggregate.

**Reading the evidence:** FACT means directly observed source data or code behavior; VENDOR means an operator's own assertion; INFERENCE means the recommended interpretation; UNVERIFIED means not established here. Repository code availability does not establish deployed bytecode equivalence. Dashboard observations are attributable measurements, not an independent audit of their indexers.

## How quote updates reach the chain (builder ordering, Priority Update Registry, signed updates, relays)

FACT/VENDOR, checked 2026-10-26: a maker computes prices offchain, signs update transactions and sends replacements privately to a builder. The builder simulates trades against its latest received state and places the applicable update before consumers. Conditional inclusion avoids publishing updates without a fill; unconditional updates instead pay for inclusion regardless of trading. These are builder policies, not Ethereum consensus guarantees. Titan's maker interface uses block-targeted signed transaction bytes, replacement identifiers and increasing replacement sequences over an authenticated WebSocket. Cancellation replaces the outstanding message with an empty transaction. [Titan maker documentation][TM], retrieved 2026-10-26.

The Priority Update Registry (PUR) limits writes to target-scoped state lanes. Direct updaters require target authorization. Signed batches use EIP-712; authorized ECDSA updaters or target-contract ERC-1271 signatures are supported, and anyone may relay a batch. The signature domain binds the registry and chain. The current implementation rejects older timestamps but permits identical signed updates to be replayed while valid: there is no universal consumed-message nonce. Reading contracts supply their own acceptable timestamp bounds. [PUR README][PR] and [implementation][PC], revision dated 2026-09-16, retrieved 2026-10-26.

FACT: the Factory consumer requests a registry timestamp equal to the current block timestamp; missing current-block state therefore fails closed. Its pair pricing is computed from published parameters and vault liquidity. [Factory contract][FC], revision dated 2026-10-23, retrieved 2026-10-26. This does not prove every proprietary venue uses the same freshness rule.

BuilderNet documents conditional inclusion by registered fill-event address/topic, and a taker holding period of **50 ms**. Update transactions have no priority fee, but makers pay a separate volume-based charge whose public amount is unavailable. Its state snapshot RPC can be used with `eth_call` state overrides. [BuilderNet API, “Priority updates” and quote-stream sections][BN], retrieved 2026-10-26. A quote stream is a simulation input; it does not reserve a price.

INFERENCE: distinguish the maker-to-builder transport, a signer-independent PUR batch relayer, and the MEV-Boost builder-to-proposer relay. None makes the others unnecessary. A proposer relay delivers a built block; it does not independently enforce this application's latest-quote promise. Budget and monitor maker delivery and taker delivery independently.

## Active venues and operators (El Zorro/Zonda, others)

VENDOR: LambdaClass identifies **El Zorro as its collaboration with Zonda Markets**, and says it runs the open-source Factory stack in production. [Factory announcement, 2026-10-14][FA]. This establishes the publicly claimed operator relationship, not custody or beneficial ownership of every associated wallet.

FACT: the pinned Router README and Titan's current taker interface include the venues below. The former is repository configuration, not a live allowlist read. The newer list includes Stelaxis and Goliath; the earlier cached Router README's shorter list and older addresses are outdated. In particular, distinguish pool/router addresses from oracle/stream addresses. [Router README][RR], revision 2026-10-23; [Titan taker interface][TT], retrieved 2026-10-26.

**Venue volume, 30 complete UTC days: 2026-09-25/2026-10-25 (end exclusive).** USD volume and decoded swap-event counts from [pamm.wtf `/api/v1/volume/daily`][PV], retrieved 2026-10-26. These counts are **not verified distinct transaction counts**; the CSV deliberately leaves `tx_count` empty.

| Venue | Volume USD | Decoded swaps | Operator attribution | Source; as of |
|---|---:|---:|---|---|
| FermiSwap | 1379286270.66 | 58675 | Operator identity UNVERIFIED | [daily API][PV]; 2026-10-26 |
| Metric | 621708330.20 | 58287 | Metric venue; legal operator UNVERIFIED | [daily API][PV]; 2026-10-26 |
| Kipseli | 51693060.73 | 10359 | Operator identity UNVERIFIED | [daily API][PV]; 2026-10-26 |
| Tempest | 47403411.91 | 13182 | Operator identity UNVERIFIED | [daily API][PV]; 2026-10-26 |
| GoliathAMM | 32692687.57 | 12035 | Operator identity UNVERIFIED | [daily API][PV]; 2026-10-26 |
| Stelaxis | 25209439.02 | 9768 | Operator identity UNVERIFIED | [daily API][PV]; 2026-10-26 |
| bopAMM | 17492722.66 | 5194 | Bebop-linked venue, separate from RFQ fills | [daily API][PV]; 2026-10-26 |
| El Zorro | 14942550.71 | 13539 | Zonda Markets / LambdaClass collaboration | [daily API][PV], [announcement][FA]; 2026-10-26 |
| TaurusFi | 5663587.11 | 3266 | Operator identity UNVERIFIED | [daily API][PV]; 2026-10-26 |
| Tracked venues total | 2196092060.57 | — | Coverage is the dashboard's observed set | [daily API][PV]; 2026-10-26 |

Hanji appears in the API's protocol metadata but has no observations in this interval; it is not assigned zero volume or asserted active. A name in metadata is insufficient evidence of current trading. Venue totals overlap Router-routed flow; do not add Router volume to them.

FACT/VENDOR comparator: WOOFi documents an oracle-priced pool with a common quote token and multiple strategist-added base tokens, plus direct pool and router integration paths. Its listed deployments do not establish an Ethereum mainnet sPMM in the sources inspected. **Ethereum WOOFi volume, update gas and swap gas: UNVERIFIED, not zero.** [WOOFi EVM integration guide][WI] and [oracle feeds][WF], retrieved 2026-10-26. The latter uses mid-price, spread and liquidity parameters plus a Chainlink feasibility safeguard. No measured superiority over PUR venues follows from those design differences.

## Router and aggregator access (LambdaClass Router and alternatives) with volumes

FACT: LambdaClass publishes Router proxy `0x4ddf368080cd7946db5b459ad591c350158175e1`, onchain venue selection and Uniswap V3 fallback. Current repository code also documents split routing and capacity-aware `quoteFillable`; do not infer that these later features were covered by the earlier audit or are deployed merely because they are on `main`. [Pinned Router README][RR], revision 2026-10-23, retrieved 2026-10-26.

VENDOR: the audit announcement reports approximately **13000000 USD/day and 1000 swaps/day as of 2026-09-22**. The averaging interval, independent query, split between proprietary fills and fallback, and current rate are **UNVERIFIED**. The source reports an audited upgrade on **2026-09-18**; that claim does not audit the maker pricing engines. [Audit announcement, 2026-09-22][RA]. The approximate daily rate is retained separately in the CSV with `estimate=true`; its bounds are explicitly unknown.

| Access choice | Engineering consequence | Attributable Ethereum volume |
|---|---|---|
| LambdaClass Router / SDK | Fresh-state simulation, inclusion-time selection, minimum output and fallback; measure the gas of scanning and split execution | Vendor daily rate above; current route-specific volume UNVERIFIED |
| Titan candidate bundles | Supply alternative AMM/PropAMM paths; builder evaluates candidates against current maker state. Stream access can be maker-restricted | Route-specific volume UNVERIFIED; venue table is not an aggregator breakdown |
| Direct pool adapter / solver integration | Avoid some Router overhead, but own token transfers, capacity checks, fallback and simulation freshness | Per-integration volume UNVERIFIED |
| WOOFi direct pool vs WOOFi router | Existing public guide identifies direct integration as slightly more gas efficient; no Ethereum apples-to-apples benchmark | Ethereum deployment/volume UNVERIFIED |

Technical basis: [Titan taker documentation][TT], [Router README][RR] and [WOOFi guide][WI], checked 2026-10-26. No live wallet or named aggregator integration is asserted from SDK availability alone.

**Addressable pair classes.** The following are author sums of the public [Blockworks query 4770 output][BW], **2026-09-25/2026-10-25 UTC**, retrieved 2026-10-26. Shares use this dataset's own total, never a DefiLlama denominator.

| Pair class | Volume USD | Share of measured Ethereum DEX volume % | Source; as of |
|---|---:|---:|---|
| ETH/stables | 7538070871.67 | 24.136225 | [query 4770][BW]; 2026-10-26 |
| BTC/stables | 1494952688.49 | 4.786704 | [query 4770][BW]; 2026-10-26 |
| LST | 1036791329.57 | 3.319713 | [query 4770][BW]; 2026-10-26 |
| stable/stable | 17093191125.32 | 54.730861 | [query 4770][BW]; 2026-10-26 |
| long tail, defined below | 2266130500.97 | 7.255946 | [query 4770][BW]; 2026-10-26 |
| Other foreign tokens, excluding BTC/stables | 1802221395.59 | 5.770551 | [query 4770][BW]; 2026-10-26 |
| Total | 31231357911.62 | 100 | [query 4770][BW]; 2026-10-26 |

Classification is explicit rather than inferred from token symbols: ETH/stables = `category_nativestablecoin_volume_usd`; BTC/stables = `pair_bitcoin_usd_volume_usd`; LST = `category_lstswaps_volume_usd`; stable/stable = `category_stablecoinswaps_volume_usd`. “Long tail” is an analytical basket summing project tokens, memes, composites and tokenized assets; it does not imply all are illiquid. The remaining foreign-token category includes BTC/ETH and other correlated or bridged-asset trades. It is kept separate so BTC/ETH is not mislabeled BTC/stables or long tail. LST follows the provider's taxonomy and may include restaking derivatives. These categories reconcile to the total before rounding; the underlying token-address taxonomy and wash-trade filters were not independently audited. [Dashboard][BD] and [category methodology][BM], checked 2026-10-26.

**Source conflict:** for the identical calendar interval, DefiLlama's Ethereum daily series sums to **37760514909.42 USD**, versus Blockworks' **31231357911.62 USD**. These are conflicting totals with unreconciled coverage/valuation differences, not interchangeable denominators. [DefiLlama daily API][LD] and [Blockworks][BW], retrieved 2026-10-26. PropAMM rows leave Ethereum shares empty because a matching, verified complete coverage denominator is unavailable. Do not silently divide their sum by either total. The brief's “largest chain by DEX volume” is also not supported for this window: Solana's same-window DefiLlama series sums to **63418237507.89 USD**, exceeding Ethereum's figure. [Solana daily API][LS], retrieved 2026-10-26.

INFERENCE: stable/stable dominates addressable gross notional but can have a very small pricing edge, so compare net output after gas, inventory financing and builder fees. ETH/stables and BTC/stables are the better initial tests of fast repricing; LST expansion needs redemption-rate and depeg models. A long-tail launch needs evidence of hedgeable inventory and reliable feeds, not just a large token list.

## Gas cost per quote update and per swap, sourced or marked UNVERIFIED

**Direct receipt observation:** in Ethereum block **26275449**, an El Zorro-targeted PUR `updateState` transaction at index **3** preceded a transaction containing an El Zorro fill at index **4**, both timestamped **2026-10-26T07:44:35Z**. Decoded update calldata contains **3 state words**. Receipts were fetched with `eth_getTransactionReceipt`, block transaction ordering with `eth_getBlockByNumber`, from [PublicNode RPC][RPC]. This demonstrates one instance of ordering, not a universal builder guarantee. [Update transaction][UT], [fill-containing transaction][ET]; checked 2026-10-26.

| Observation / operation | Gas units | Effective gas price gwei | Approximate execution fee USD | Source; as of |
|---|---:|---:|---:|---|
| El Zorro PUR update, isolated transaction | 41986 | 0.045042449 | 0.00438673 | [receipt][UT]; 2026-10-26 |
| Whole transaction containing El Zorro swap | 1270043 | 0.095042449 | 0.27999514 | [receipt][ET]; 2026-10-26 |
| Whole transaction containing FermiSwap swap | 879547 | 0.043615507 | 0.08898451 | [receipt][FT]; 2026-10-26 |
| Whole transaction containing Metric swap | 334379 | 0.111110826 | 0.08618066 | [receipt][MT]; 2026-10-26 |
| Whole transaction containing bopAMM swap | 417956 | 0.043365155 | 0.04204225 | [receipt][BT]; 2026-10-26 |

USD conversion uses **2319.607222109381 USD/ETH**, [DefiLlama price endpoint][EP], observation timestamp **2026-10-26T07:45:01Z**. Formula: `gasUsed × effectiveGasPriceWei × ETHUSD / 10^18`. USD fees are estimates marked `estimate=true`; gas and effective gas price are exact receipt fields. These transactions are convenience samples selected by the dashboard's latest venue fill, not randomly sampled or a venue ranking. Receipts include other execution; swap events in one transaction share its receipt gas. Never treat these rows as the isolated cost of a venue swap or sum receipt gas across its logs. Explicit builder transfers, separate update transactions and offchain volume-based charges are excluded.

**Published benchmarks, not independently rerun or verified mainnet costs:**

| Benchmark | Gas units | Interpretation | Source; revision date |
|---|---:|---|---|
| PUR direct update `k=0` | 30698 | Packed first word only | [snapshot][PG]; 2026-09-16 |
| PUR direct update `k=5` | 60882 | Additional words beyond the first | [snapshot][PG]; 2026-09-16 |
| PUR direct update `k=10` | 91311 | Larger payload | [snapshot][PG]; 2026-09-16 |
| PUR signed batch `n=5,k=0` | 116008 | Whole signed batch, not each update | [snapshot][PG]; 2026-09-16 |
| PUR signed batch `n=10,k=0` | 207802 | Whole signed batch | [snapshot][PG]; 2026-09-16 |
| Factory cold `quote` | 38890 | Mock registry/token environment | [snapshot][FG]; 2026-10-23 |
| Factory warm `quote` | 8890 | Warm call, not an update | [snapshot][FG]; 2026-10-23 |
| Factory `quoteFillable` | 44216 | Capacity plus output calculation | [snapshot][FG]; 2026-10-23 |
| Factory warm swap leg | 65467 | Measured after capacity quote and token push | [snapshot][FG]; 2026-10-23 |

PUR `k` counts additional words; `k=0` is not an empty price update. Tests pre-initialize storage, so do not extrapolate to first writes. Factory tests explicitly warn that mock PUR read cost differs from real PUR; its swap snapshot excludes preceding quote/push and is not an end-to-end Router transaction. [PUR benchmark test][PT], revision 2026-09-16; [Factory gas test][GT], revision 2026-10-23. CSV benchmark USD/gas-price cells are empty: the tests state neither. Typical isolated mainnet swap cost remains **UNVERIFIED**.

INFERENCE: offchain pricing avoids putting exchange feeds and the full pricing model in the EVM, but transfers, signature checks, storage and settlement remain. Conditional inclusion amortizes actual updates across filled flow. A shared signed batch can amortize transaction overhead across makers; it also adds signature work. For example, the published packed-word batch snapshots imply **23201.6 gas/update** at **5 updates** and **20780.2 gas/update** at **10 updates**, dividing whole-batch gas by batch size; these derived averages are marked `estimate=true`, not mainnet measurements. [PUR snapshots][PG], revision 2026-09-16, retrieved 2026-10-26. Candidate-route bundles primarily improve price selection offchain; they do not make each executed swap free. Onchain requoting consumes gas. Benchmark direct execution, routed execution, split execution, fallback and failed fills separately before selecting an integration.

## Dependence on block builders and the centralisation risk

FACT/VENDOR: Titan's current maker rules use configurable freshness protection: a taker must have arrived earlier than the update by the configured buffer. The documented example buffer and recommended refresh cadence are **50 ms**; streams evict quotes not refreshed for **400 ms**. Maker API access is permissioned. [Titan makers][TM], checked 2026-10-26. BuilderNet describes a minimum taker holding period rather than identically phrased per-update timing. Do not assume all builders enforce identical semantics. [BuilderNet][BN], checked 2026-10-26.

INFERENCE: this grants a builder control over the effective trading venue: update selection, taker eligibility, routing visibility and inclusion. A successful block auction does not prove the taker received neutral sequencing. Signed updates authenticate the maker, not the builder's completeness or fairness. Permissioned streams and private volume pricing create access and commercial concentration risks even when the pool's swap entrypoint is public. Measure filled volume, failed fills, stale-state rejects, quote-to-fill deterioration and maker markouts by winning builder, pair, size and route. Track tail losses as well as medians.

A multi-builder adapter reduces transport dependence, but does not guarantee diverse winning block production or uniform last-look rules. Source-based context: [Neuder/Bahrani, Ethereum Research, 2026-10-07][ER] discusses unequal execution access and maker/builder integration incentives. Current Ethereum builder market shares and causal builder-specific execution advantages are **UNVERIFIED** here; Solana leader-slot markouts cannot supply those missing estimates.

## Spoofing and requote protections

FACT: the Router offers inclusion-time venue comparison, a caller minimum output and fallback; these constrain a venue that attracts simulated flow then worsens its executable price. They do not guarantee the original displayed quote, neutral counterparty treatment, or execution better than every offchain alternative. [Router documentation][RR], revision 2026-10-23. INFERENCE: compare simulation and fill under the same sender, caller, size, token path and fee treatment. A minimum output that tolerates deterioration still permits deterioration. Prefer measured recipient balance deltas, bounded quote age, token-specific circuit breakers and price-deviation limits; test fee-bearing fallback and malicious venue behavior.

**Material prior incident:** Flashbots' advisory says the original registry allowed lane writes to alias updater authorization storage, enabling arbitrary target prices or updater revocation. The corrected mainnet address is `0xDa7AfEeD021EAFC1c1Af9C362dE477DaD0396B81`; the superseded `0xDa7AfeeD01fe625CF15d187a19f94B45f00b8C5F` remains vulnerable. Migration requires reauthorizing updaters and reinitializing lanes, not just replacing an address. The advisory's lack of observed exploitation is explicitly limited evidence. [Flashbots advisory, 2026-09-16][SA]. INFERENCE: independently constrain acceptable prices at settlement, even with an authenticated registry; test domain changes, updater revocation and replay behavior. Keep quoting keys separate from inventory authority.

## What it takes for a new market maker to supply liquidity here

Recommended build sequence (engineering judgment):

- Implement a small adapter around the existing pricing engine: pair-specific feed normalization, inventory skew, validity, signer rotation and builder-specific replacement/cancellation. Compare its net execution and inventory losses against the existing WOOFi path under matched conditions.
- Choose a settlement contract and custody model. The Factory's documented instance uses per-pair vaults with token allowances; that is not automatically equivalent to WOOFi's pool inventory design. Review authorization, emergency pause, token accounting and upgrade rights before moving funds.
- Arrange maker API access with builders, negotiate per-pair commercial fees, register consumption events where needed, and operate publishers near the selected endpoints. Capital and builder access are not supplied by an open-source repository.
- Integrate both fresh-state simulation and executable settlement. Add a Router adapter for broad venue access, plus a direct adapter for solvers willing to own routing/fallback. Publish explicit pair discovery, capacity, freshness and failure behavior; SDK publication alone will not secure distribution.
- Instrument delivery acknowledgments, quote landings, inventory/allowances, fill markouts, failed trades and all-in cost. Launch further assets or chains only after showing adequate net pricing and reliable data/hedging for those assets.

FACT basis: [Factory production guide][FP], revision 2026-10-23, documents deployment, inventory approval, funded per-pair updating wallets and separate builder onboarding. Its [components][COMP] expose capacity discovery; its [pricing guide][PRICE] supplies feed, fixed and inventory/volatility-responsive models. These are usable scaffolding, not evidence of a profitable strategy. WOOFi's [integration guide][WI], checked 2026-10-26, supplies a comparable pool-versus-router design choice; neither provider has a matched mainnet gas/execution study in the evidence gathered.

## Open questions

**Resolve before a production decision:** what are net spreads after per-pair builder charges, hedge costs, funding and failed transactions; which builders actually land this maker's flow; how do freshness guarantees differ under outages; which router implementation and features are currently deployed and audited; what portion of venue notional is aggregator/wallet demand rather than arbitrage; how much long-tail notional survives independent token/feed and wash-trade checks? All remain UNVERIFIED. They require public measurements or commercial onboarding, not non-public information about the requesting team.

**Starting-lead verification:**

- The arXiv paper exists, submitted **2026-09-29**. Its data-collection section/footnote excludes Ethereum and characterizes its builder model as closer to RFQ. It reports Base's start as **2025-10-30** and Monad's first PropAMM in early **2026-04**. Its sample ends **2026-08-31**, so its provider shares are not a later rolling-month estimate. The named Solana providers and Base Tessera/Metric/Elfomo appear, but “first and largest” remains a statement about the paper's observed sample, not a permanent ranking. [Paper, sections 4–5 and appendix A][PAPER]. No Solana or Base share is imported into the Ethereum CSV.
- The TDMM/DefiLlama rolling-month lead to **2026-09-26** is **UNVERIFIED** here; no such numerical series is used. A paper-period share, a selected-pair share, an average daily share and a ratio of total monthly notional are different statistics. Reconciliation requires the same dates, asset universe and weighting; averaging them would be invalid.
- The Router and Factory leads are confirmed by their repositories and announcements, but cached venue/address lists can be stale. The latest repository's extra routes should be checked against deployed bytecode before integration. [Router][RR], [Factory][FA], checked 2026-10-26.
- The supplied Blockworks report date is wrong for the matching title: **“Solana’s Block Building Wars” is dated 2026-02-17**, not 2026-10-14. Its leader-slot markout study is Solana-specific. A separate newsletter has a similar title. [Original report][WARS], checked 2026-10-26. It motivates builder-stratified measurement; it does not establish Ethereum results.

**Reproduction and CSV contract:** `ethereum-volume.csv` uses USD, one header, long-format daily venue/category rows and aggregate rows. Select either daily or aggregate windows, not both. Calendar windows are ISO start/end intervals, end exclusive. The sole vendor rate uses an ISO duration; its venue label contains the ISO reporting date and explicitly unknown bounds. `as_of_date` means retrieval date; benchmark revision dates and transaction timestamps are stated above. The added `estimate` column satisfies the assignment's estimate-label rule while retaining all requested columns. False means no estimation beyond aggregation/rounding of the stated source; it is not a certification of indexer accuracy. Missing data is empty, not zero. Source `swaps` and `total_trade_cnt` were not relabeled as distinct transactions, so `tx_count` remains empty. Router daily swap count is prose-only for the same reason.

The market table can be re-pulled from **Blockworks dashboard 938, visualization 9331, query 4770**, execution `airflow-8c1bf874-6c42-4eae-a872-acfb4152735b`. Filter `date >= 2026-09-25T00:00:00Z AND date < 2026-10-25T00:00:00Z`, then sum the named columns. Cross-check categories with **visualization 9320/query 4766**, execution `airflow-869430b9-d8c6-4892-a0cb-4751ac5f8e99`, via its [public endpoint][BC]. Both returned complete result counts; the chosen interval contains complete daily rows. BTC/stables is query 4770's explicit BTC-USD field, not all foreign-token volume. The exact API and query IDs are supplied instead of digitizing chart pixels.

For venue data use [the daily API][PV], filter its `day` by that same interval and group `protocol_code`; sum `volume_usd`. This endpoint has no pair breakdown, so `pair=ALL` is intentional. For DefiLlama use [the daily series][LD], convert timestamps to UTC dates, and sum the identical interval; do not substitute its live `total30d` field. For gas, use the pinned JSON snapshot keys in the CSV, or fetch each cited transaction receipt and compare `gasUsed`/`effectiveGasPrice`; verify the update selector against the pinned PUR ABI. USD conversions use the separate cited ETH/USD observation and exclude commercial builder fees. All transformations can be repeated without an account using the cited public data interfaces while available; historical indexer revisions can change later results.

**Local checks:** CSV schemas, finite nonnegative numeric fields, ISO observation dates, complete market daily coverage, category/venue aggregate reconciliation, receipt fields and USD conversion were checked locally. No contract build was needed for this research deliverable. Source truth, deployed-code equivalence and profitability have not been certified by these checks.

[BW]: https://app.blockworks.com/api/blockworks-studio/dashboard/938/visualization/9331/execution
[BC]: https://app.blockworks.com/api/blockworks-studio/dashboard/938/visualization/9320/execution
[BD]: https://app.blockworks.com/projects/ethereum/analytics/ethereum-spot-dexs
[BM]: https://blockworks.co/analytics/ethereum/ethereum-dex
[PV]: https://pamm.wtf/api/v1/volume/daily
[LD]: https://api.llama.fi/overview/dexs/Ethereum?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=true
[LS]: https://api.llama.fi/overview/dexs/Solana?excludeTotalDataChart=false&excludeTotalDataChartBreakdown=true
[TM]: https://docs.titanbuilder.xyz/propamms/makers
[TT]: https://docs.titanbuilder.xyz/propamms/takers
[BN]: https://buildernet.org/docs/api
[PR]: https://github.com/flashbots/priority-update-registry/blob/8701e8e13d74beef9126c3e1fbbe89c64ce0c0f6/README.md
[PC]: https://github.com/flashbots/priority-update-registry/blob/8701e8e13d74beef9126c3e1fbbe89c64ce0c0f6/src/PrioUpdateRegistry.sol
[PG]: https://github.com/flashbots/priority-update-registry/blob/8701e8e13d74beef9126c3e1fbbe89c64ce0c0f6/snapshots/GasBenchmarkTest.json
[PT]: https://github.com/flashbots/priority-update-registry/blob/8701e8e13d74beef9126c3e1fbbe89c64ce0c0f6/test/GasBenchmark.t.sol
[FA]: https://blog.lambdaclass.com/announcing-propamm-factory/
[FC]: https://github.com/lambdaclass/propamm-factory/blob/8ded63cd09dc6ee628b17188a13ced14b24a7ea1/contracts/src/PropAMM.sol
[FG]: https://github.com/lambdaclass/propamm-factory/blob/8ded63cd09dc6ee628b17188a13ced14b24a7ea1/contracts/snapshots/PropAMMGasTest.json
[GT]: https://github.com/lambdaclass/propamm-factory/blob/8ded63cd09dc6ee628b17188a13ced14b24a7ea1/contracts/test/PropAMMGas.t.sol
[FP]: https://github.com/lambdaclass/propamm-factory/blob/8ded63cd09dc6ee628b17188a13ced14b24a7ea1/docs/production.md
[COMP]: https://github.com/lambdaclass/propamm-factory/blob/8ded63cd09dc6ee628b17188a13ced14b24a7ea1/docs/components.md
[PRICE]: https://github.com/lambdaclass/propamm-factory/blob/8ded63cd09dc6ee628b17188a13ced14b24a7ea1/docs/pricing.md
[RR]: https://github.com/lambdaclass/propamm-router-contracts/blob/eb8adc7df98157180def9b41802f95972173fe4a/README.md
[RA]: https://blog.lambdaclass.com/lambdaclass-propamm-router-audited-version-now-live/
[WI]: https://learn.woo.org/woofi-docs/woofi-dev-docs/guides/integrate-woofi-as-liquidity-source
[WF]: https://learn.woo.org/woofi-docs/woofi-dev-docs/resources/on-chain-price-feeds
[ER]: https://ethresear.ch/t/ethereum-propamms-today-and-tomorrow/26129
[SA]: https://collective.flashbots.net/t/security-advisory-storage-aliasing-in-priority-update-registry-v1-09-15-26/6030
[PAPER]: https://arxiv.org/html/2609.38056v1
[WARS]: https://app.blockworks.com/report/solana-s-block-building-wars
[RPC]: https://ethereum-rpc.publicnode.com
[EP]: https://coins.llama.fi/prices/current/coingecko:ethereum
[UT]: https://etherscan.io/tx/0xdcfdc0bf284032d96e637ccd875a82a41eb86e9fa5614ac24dabb37d783b636f
[ET]: https://etherscan.io/tx/0x678d50511f7a00703790b28887b05ee5fbcc571cbfc53de1f583a36494243cc2
[FT]: https://etherscan.io/tx/0x72f78602055657b63c8eb7a99bf44c172a6954d91648e8db0efe551b182c05ba
[MT]: https://etherscan.io/tx/0x5137548a77a7440af23360e49f8471f9bcf4fcb2d1d418c7c8f3823c0f38c3ab
[BT]: https://etherscan.io/tx/0x2d28eb3aa2b6fb7d8bdfaf1fe04bfc4cb4f9954789fc60e98cab4d0d987f0488