Pith. sign in

REVIEW 5 major objections 4 minor 17 references

A Coincidence of Wants Mechanism for Swap Trade Execution in Decentralized Exchanges

T0 review · 5 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read This paper claims that incomplete Coincidence of Wants swap cycles on decentralized exchanges can be completed by adding a single synthetic bridging order derived from a matrix imbalance, yielding slippage-free, capital-preserving…

desk verdict A tidy matrix formalization of CoW cycles whose key 'bridging order' is a synthetic order with no counterparty, so the claimed value-neutral completion is an accounting identity, not a trade. read the letter →

arxiv 2507.10149 v1 pith:66QP5H4N submitted 2025-07-14 cs.GT cs.CEq-fin.TR

classification cs.GTcs.CEq-fin.TR MSC 91B2605C2068R10
keywords CoincidenceofWantsdecentralizedexchangeswapexecutionassettransfermatrixbridgingordervalueneutralitypartialfillscycledetection
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that Coincidence of Wants (CoW) cycles—closed loops of swap orders where each trader's requested asset is the next trader's offered asset—can be detected and, when incomplete, completed by one synthetic bridging order derived from the system's matrix imbalance. If true, DEX aggregators could settle trades without tapping AMM liquidity pools, avoiding slippage, and liquidity providers would end each cycle with exactly the inventory they started with. The authors encode swap flows in an asset transfer matrix, use oracle prices to convert to dollar terms, and show that closing a cycle amounts to making both row and column sums of that matrix zero. Applying the algorithm to ten real Arbitrum swaps, they find one naturally closed three-asset cycle and propose small bridging orders to close two others.

What carries the argument

The central object is the asset transfer matrix $M$ and its dollar-normalized version $V = M \cdot P$, where each row is a swap order, each column is an asset vault, a positive entry denotes an asset offered, and a negative entry denotes an asset requested. The imbalance vector $N = \mathbf{1}^T V$ records the per-asset net dollar flow, and a complete cycle has $N = 0$. A bridging order is then the row $-N$ that restores column-sum balance, turning cycle completion into a linear algebra operation. Feasibility is enforced by requiring each trader's realized exchange rate to be at least the oracle reference rate.

What would settle it

Take the paper's Arbitrum example: the ETH to ARB to USDC to ETH cycle is scaled to $0.24 per leg, so if the trader who submitted the $2442.17 ARB to USDC order declines a $0.24 partial fill, no complete cycle exists. A concrete test is to run the algorithm on a public order batch and check whether any counterparty voluntarily submits the proposed bridging orders (UNI to aArbWETH and USDC to USDT) at the oracle price within the batch window.

Watch

Extended reading notes

Core claim

The central claim is that any partial collection of swap orders can be completed into a value-neutral CoW cycle exactly when a single synthetic order, the bridging order, is added whose dollar flow vector is the negative of the system's imbalance vector $N = \mathbf{1}^T V$. With that order added, the dollar-normalized transfer matrix has zero row and column sums, meaning every trader receives dollar-equivalent value for what they give and every asset's total supply equals total demand. The paper calls this 'slippage free and capital preserving for LPs' and presents it as provable value neutrality, in contrast to the heuristic matching used by existing batch-auction solvers.

Load-bearing premise

The paper's guarantees rest on the assumption that someone will actually supply the one missing trade the math says is needed, at the reference price, and that all traders will accept very small partial trades.

Editorial extensions

If this is right

  • DEX aggregators could execute batches of swap orders entirely through internal matching, without accessing an AMM's liquidity pool, whenever a CoW cycle exists.
  • Incomplete cycles can be closed atomically with one synthetic order, and the paper argues this preserves LP capital because each vault ends the cycle with the same inventory it started with.
  • The framework allows partial fills scaled to the smallest dollar value in the cycle, so even very small trades, such as the $0.24 example on Arbitrum, can form feasible cycles.
  • Using oracle prices to anchor value neutrality makes the resulting execution resistant to slippage and to value-extractive settlements against imbalanced prices, mitigating MEV risk.
  • Cycle enumeration via Johnson's algorithm has exponential worst-case complexity, but the paper argues it stays tractable in practice for small transaction batches and sparse asset graphs.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The bridging order is a theoretical construct: the paper does not specify who supplies it or why a counterparty would accept the trade at the oracle price, so a real deployment would need an incentive mechanism or solver competition to fill it.
  • The 'capital preserving' guarantee holds only if the bridging order actually fills; otherwise the LP vault providing the missing leg carries the imbalance, making the neutrality claim contingent on complete atomic settlement.
  • One testable extension is to run the imbalance-vector construction on live order flow and measure how often a proposed bridging order is voluntarily taken by market participants, which would reveal whether the mechanism works beyond the paper's simulated examples.
  • If bridging fills reliably, the resulting cycles are delta-neutral for LP vaults, a property the paper itself notes as a potential strategy for liquidity-providing market makers.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 4 minor

Summary. The paper proposes a matrix-based framework for detecting and completing coincidence-of-wants (CoW) cycles among DEX swap orders. Swap orders are represented as rows of an asset transfer matrix, converted to dollar values via an oracle price vector, and partial cycles are closed by adding a synthetic 'bridging order' whose flow is the negative of the column-sum imbalance vector. The authors claim that this construction yields slippage-free, value-neutral, LP-capital-preserving execution, and they illustrate the approach on ten Arbitrum swap transactions, reporting complete cycles and bridged cycles.

Significance. If the central claims were correct, a formal graph-plus-matrix characterization of CoW cycles with a constructive completion procedure would be a useful addition to the batch-auction DEX literature, and the bounded-cycle complexity analysis would give a reasonable starting point for practical implementations. The use of real Arbitrum swap data is also a step toward empirical evaluation. However, the significance is contingent on the bridging order being economically executable and on the matrix conservation statements being correct; as detailed below, the central construction is an algebraic identity rather than a market outcome, and the formal derivation contains sign and well-definedness errors that invalidate the claimed guarantees.

major comments (5)
  1. [Section 3.4] The claim that quantity mismatch can produce an aggregate surplus or deficit with product of realized exchange rates different from 1 is internally inconsistent: with r̂_{i_j} := q_{j+1}/q_j and q_{k+1} := q_1, the product telescopes to exactly 1 by definition. The subsequent discussion of 'net surplus (arbitrage opportunity) or deficit (loss)' therefore has no mathematical basis as stated.
  2. [Section 4.1.1 and Section 4.2] The bridging-order derivation contains sign errors. The imbalance is computed as N = [3000, 0, -3000], but the text then writes Vnew = -N = [+3000, 0, -3000]; the correct negative is [-3000, 0, +3000]. Consequently Mnew = [+1, 0, -1500] corresponds to a swap ETH -> ARB, not the claimed order (ARB, ETH, a3=1500, b3=1). The displayed matrices M' and V' also have their row/column sum statements reversed: the displayed M' has zero column sums and nonzero row sums, not 'only the row sums are zero' as asserted. Because Definition 8 and Observation 4 rest on this example, the formal derivation of the bridging order is invalid as written.
  3. [Definition 8 and Algorithm 4.3] The bridging order defined as V_{k+1} := -N is not generally a valid swap order under Definition 1. A swap order has exactly one input asset and one output asset, so if the imbalance vector -N has nonzero entries in more than two asset columns, no single swap order can represent it. Even when -N has support on exactly two assets, the definition supplies no counterparty, no limit price, and no incentive mechanism guaranteeing that anyone will take the synthetic leg; the identity 1^T V' = 0 holds by construction, not as a market-clearing outcome. The paper's claims of 'value-neutral execution' and 'capital preserving for LPs' therefore do not follow: an LP that fills the bridge changes its inventory unless some unmodeled market-making or compensation mechanism is introduced.
  4. [Section 5.2 and Section 5.3] The empirical evaluation assumes traders accept arbitrarily small partial fills. In Section 5.2 a cycle containing orders of $2442.17 and $488.52 is scaled down to $0.24 for every leg, and in Section 5.3 an order of $20.24 is scaled to $0.34. The paper provides no basis in Definition 1 or in the data for assuming such negligible fills are acceptable, so the resulting 'cycles' are artifacts of the min-value normalization rather than demonstrated executable CoW opportunities. The bridged cycles in Sections 5.3 and 5.4 are closed by synthetic orders that do not appear in Table 1 and for which no taker is identified.
  5. [Section 4.3, Algorithm step 2(a)] The algorithm computes N = 1^T M_C, where M_C is the asset transfer matrix, whereas Definition 7 defines N := 1^T V for the dollar-normalized matrix. These are different objects, and step 2(c) then defines the bridging order with 'flow -N' without a conversion through the price diagonal P. The two definitions of N must be reconciled for the algorithm to be well-defined; as written, the quantity being negated is ambiguous.
minor comments (4)
  1. [Section 3.4] The sentence defining the acceptable exchange rate says the trader expects r_i · b_i units of asset B_i; from the surrounding definitions it should be r_i · a_i units of B_i in return for a_i units of A_i.
  2. [Definition 1 and Section 5] Definition 1 requires a_i, b_i ∈ R_{>0}, but Section 5 repeatedly writes orders with negative quantities, e.g., o1 = (ETH, ARB, 0.24, -0.24) in Section 5.2 and o4 = (UNI, aArbWETH, +0.34, -0.34) in Section 5.3.4; the notation should be made consistent.
  3. [Section 4.3] The complexity expression O((|V|^2 log|V| + |V||E|)(c · k_max) has an unbalanced parenthesis and should be completed, e.g., as O((|V|^2 log|V| + |V||E|)(c + 1) · k_max).
  4. [Section 1] The introduction promises a 'formal proof' in Section 4, but the paper provides observations and worked examples rather than a proof environment; the presentation should either include the proof or soften the claim.

Circularity Check

2 steps flagged · score 7.0 of 10

Core 'value-neutral execution' claim is definitional: bridging orders are defined as the negative imbalance vector, so conservation is an identity, and the real-data 'cycles' are balanced by min-fill normalization.

  1. self definitional [Section 4.2.2, Definition 7, Definition 8, Observation 4]
    "Define a bridging order ok+1 with net flow vector Vk+1 := −N so that: V ′ := ( V −N ) ⇒ 1⊤V ′ = 0. This enforces closure of the cycle in asset space. ... The vector −N corresponds to a unique bridging order up to scaling that restores global asset neutrality."

    N is defined as 1^T V in Definition 7, so appending a row equal to −N makes 1^T V' = 0 by elementary construction, not by any market outcome. The claimed 'closure', 'value neutrality', 'slippage free' execution, and 'capital preserving for LPs' therefore follow from the definition of V', not from the existence of a trader willing to take the synthetic bridge. The paper's own conclusion concedes that incentives for synthetic bridging remain future work, confirming that the bridging order is an algebraic placeholder rather than a discovered want.

  2. fitted input called prediction [Section 5.2, 'Dollar-Normalized Matrix']
    "Given the lowest dollar value is $0.24, this would correspond to a partial order fill across all orders in the cycle. Therefore the matrix V would look like the following: M = [[+0.24, -0.24, 0], [0, +0.24, -0.24], [-0.24, 0, +0.24]]"

    The real-data demonstration creates value balance by scaling every order to the smallest USD amount, $0.24. Each row then has identical +0.24/−0.24 entries, so both row sums and column sums vanish automatically. The 'efficient discovery of CoW cycles' is thus produced by the chosen fill normalization and assumed oracle prices, not inferred from the orders as recorded; the feasibility test is satisfied by construction.

full rationale

The paper's formal machinery is internally coherent, but its central guarantee is vacuously definitional. In Definition 8, the bridging order is literally set to the negative of the imbalance vector N = 1^T V, so conservation 1^T V' = 0 is an algebraic identity. The paper then presents this identity as proof that partial CoW cycles can be completed and executed slippage-free and capital-preserving for LPs. No mechanism ensures that any trader wants the synthetic bridging leg, and in the data examples the bridges UNI→aArbWETH and USDC→USDT appear nowhere in Table 1. The empirical demonstration in Section 5.2 is similarly constructed: orders are partial-filled at the minimum dollar value, making the matrix balance by normalization rather than by discovery. There is no problematic self-citation chain; the circularity is definitional. This warrants a score of 7 rather than higher because the linear algebra itself is correct, but the load-bearing claim of executable, value-neutral completion reduces to the definition of the bridging order.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The central claim rests on two hand-selected inputs: oracle prices and a min-value partial-fill rule. It also requires an ad hoc bridging order with no demonstrated counterparty, plus strong assumptions about trader acceptance of tiny fills and passive single-asset LP vaults. These are not derived from data or prior literature.

free parameters (3)
  • Oracle price vector p = Motivating example: p(ETH)=3000, p(USDC)=1, p(ARB)=2; simulation: 1 ETH=1500 ARB=3000 USDC=1 aArbWETH=375 UNI
    Chosen by hand to make the example cycles balance; the source of oracle prices for the simulation is not specified.
  • Partial-fill notional (min dollar value in cycle) = $0.24 for Cycle 1, $0.34 for bridging example
    The algorithm scales all orders in a detected cycle to the smallest USD amount; this selection rule is what makes the sparse real data produce "feasible" cycles.
  • Cycle length cap (kmax) and batch window = kmax=3-4, batch=10 swaps within 4 minutes
    Algorithmic hyperparameters chosen for the simulation; not justified by data, and they determine which cycles are discovered.
assumptions (5)
  • domain assumption Trades in a CoW cycle execute exactly at oracle prices (value feasibility p(B_i)/p(A_i)=a_i/b_i).
    Invoked in Definition 6 and Observation 1; real orders have limit prices and may not accept oracle rates.
  • ad hoc to paper A bridging order with flow -N can always be filled at the oracle price.
    Definition 8 and Section 5.3.4; the paper supplies no counterparty or pricing mechanism for the synthetic order, only the matrix row.
  • ad hoc to paper Traders accept arbitrarily small partial fills of their orders.
    Section 5.2 fills all orders in a cycle to the minimum dollar value ($0.24); no minimum fill or trader acceptance constraint is modeled.
  • domain assumption LP vaults hold a single asset and are passive inventory providers.
    Section 3.3, Definition 3; real LPs may require fees, have multi-asset exposure, or rebalance.
  • domain assumption No fees, gas costs, or market frictions are incurred.
    Section 4.1 states "Ignoring market frictions"; this is relaxed nowhere in the main claims.
invented entities (1)
  • Bridging (synthetic) order
    purpose: A swap order with flow -N inserted to make the column sums of the dollar transfer matrix zero and close a partial CoW cycle.
    No evidence that a counterparty exists for this order; it is generated by the algorithm from the imbalance vector, not discovered from trader intents.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Coincidence of Wants Mechanism for Swap Trade Execution in Decentralized Exchanges." pith.science (2026). https://pith.science/paper/66QP5H4N

@misc{pith2026250710149,
  author       = {Pith},
  title        = {Pith review of: A Coincidence of Wants Mechanism for Swap Trade Execution in Decentralized Exchanges},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/66QP5H4N}},
  note         = {Machine review of arXiv:2507.10149}
}
read the original abstract

We propose a mathematically rigorous framework for identifying and completing Coincidence of Wants (CoW) cycles in decentralized exchange (DEX) aggregators. Unlike existing auction based systems such as CoWSwap, our approach introduces an asset matrix formulation that not only verifies feasibility using oracle prices and formal conservation laws but also completes partial CoW cycles of swap orders that are discovered using graph traversal and are settled using imbalance correction. We define bridging orders and show that the resulting execution is slippage free and capital preserving for LPs. Applied to real world Arbitrum swap data, our algorithm demonstrates efficient discovery of CoW cycles and supports the insertion of synthetic orders for atomic cycle closure. This work can be thought of as the detailing of a potential delta-neutral strategy by liquidity providing market makers: a structured CoW cycle execution.

Figures

Figures reproduced from arXiv: 2507.10149 by the authors.

Figure 1
Figure 1. A bipartite LP-resolved CoW cycle: each trader receives assets from one vault and sends [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Directed asset flow graph for a 3-order CoW cycle. Each edge denotes a desired asset [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 15 canonical work pages

  1. [1]

    The costs of swapping on decentralized exchanges

    Austin Adams, Benjamin Y Chan, Sarit Markovich, and Xin Wan. The costs of swapping on decentralized exchanges. Financial Cryptography (preprint), 2024. 19

  2. [2]

    Uniswap v3 core

    Hayden Adams, Noah Zinsmeister, Moody Salem, River Keefer, and Dan Robinson. Uniswap v3 core. Tech. rep., Uniswap, Tech. Rep., 2021

  3. [3]

    Arbitrum nitro: A second- generation optimistic rollup, 2022

    Lee Bousfield, Rachel Bousfield, Chris Buckland, Ben Burgess, Joshua Colvin, E Felten, Steven Goldfeder, Daniel Goldman, Braden Huddleston, H Kalonder, et al. Arbitrum nitro: A second- generation optimistic rollup, 2022

  4. [4]

    An analysis of intent-based markets

    Tarun Chitra, Kshitij Kulkarni, Mallesh Pai, and Theo Diamandis. An analysis of intent-based markets. arXiv preprint arXiv:2403.02525 , 2024

  5. [5]

    Sok: Decentralized finance (defi)–fundamentals, taxonomy and risks

    Krzysztof Gogol, Christian Killer, Malte Schlosser, Thomas Bocek, Burkhard Stiller, and Claudio Tessone. Sok: Decentralized finance (defi)–fundamentals, taxonomy and risks. arXiv preprint arXiv:2404.11281, 2024

  6. [6]

    Efficient algorithms for shortest paths in sparse networks

    Donald B Johnson. Efficient algorithms for shortest paths in sparse networks. Journal of the ACM (JACM), 24(1):1–13, 1977

  7. [7]

    Impermanent loss and slippage in automated market makers (amms) with constant-product formula

    Mauricio Labadie. Impermanent loss and slippage in automated market makers (amms) with constant-product formula. Available at SSRN 4053924 , 2022

  8. [8]

    A coincidence of wants: The novel and neoclassical economics

    Charles Lewis. A coincidence of wants: The novel and neoclassical economics . Routledge, 2014

Show all 17 references
  1. [9]

    Analysis and evaluation of the cow protocol for mev mitigation on the ethereum blockchain

    Vigan Lladrovci. Analysis and evaluation of the cow protocol for mev mitigation on the ethereum blockchain. Master’s thesis, Technical University of Munich, Chair of Software Engineering for Business Information Systems (sebis), Garching, Germany, 2023. Accessed: 2025-03-13

  2. [10]

    A theory of atomic trans- actions

    Nancy Lynch, Michael Merritt, William Weihl, and Alan Fekete. A theory of atomic trans- actions. In ICDT’88: 2nd International Conference on Database Theory Bruges, Belgium, August 31–September 2, 1988 Proceedings 2 , pages 41–71. Springer, 1988

  3. [11]

    Decentralized exchange

    Semyon Malamud and Marzena Rostek. Decentralized exchange. American Economic Review, 107(11):3320–3362, 2017

  4. [12]

    Maximal extractable value (mev)

    Udo Milkau. Maximal extractable value (mev). In Decentralized Finance und Tokenisierung, pages 103–108. Springer, 2023

  5. [13]

    Automated market makers and decentralized exchanges: a defi primer

    Vijay Mohan. Automated market makers and decentralized exchanges: a defi primer. Financial Innovation, 8(1):20, 2022

  6. [14]

    Balancer

    Miguel Ottina, Peter Johannes Steffensen, and Jesper Kristensen. Balancer. In Automated Market Makers: A Practical Guide to Decentralized Exchanges and Cryptocurrency Trading , pages 69–116. Springer, 2023

  7. [15]

    Competitive solvers in action: strategic optimiza- tion across cows and multi-chain amms

    Zeshun Shi, Sydney Sweck, and Omar Zaki. Competitive solvers in action: strategic optimiza- tion across cows and multi-chain amms. 2025

  8. [16]

    Decentralized finance

    Dirk A Zetzsche, Douglas W Arner, and Ross P Buckley. Decentralized finance. Journal of Financial Regulation, 6(2):172–203, 2020

  9. [17]

    Maximal extractable value in batch auctions

    Mengqian Zhang, Yuhao Li, Xinyuan Sun, Elynn Chen, and Xi Chen. Maximal extractable value in batch auctions. In Proceedings of the 26th ACM Conference on Economics and Com- putation, pages 510–510, 2025. 20

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.