Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

Measuring CEX-DEX Extracted Value and Searcher Profitability: The Darkest of the MEV Dark Forest

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read CEX-DEX arbitrage searchers extracted about $233.8M from Ethereum over 19 months, and three of them took nearly three-quarters of the value.

desk verdict The most complete empirical map yet of CEX-DEX arbitrage, but the 233.8M headline is an upper bound with an unquantified upward bias from in-sample horizon fitting; the structural concentration and profit-sharing findings are the robust core. read the letter →

arxiv 2507.13023 v3 pith:WRTWXON2 submitted 2025-07-17 cs.CR q-fin.TR

Fei Wu , Danning Sui , Thomas Thiery , Mallesh Pai This is my paper
classification cs.CRq-fin.TR
keywords EthereumMaximalExtractableValueCEX-DEXarbitrageProposer-BuilderSeparationMEV-Boostblockbuilderssearcherprofitabilitymarketcentralization
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

This paper tries to measure the largest unseen part of Ethereum's MEV economy: arbitrage between centralized and decentralized exchanges. Because only the on-chain leg is visible, the authors build a detection rule set and a markout-based revenue estimator that infers when each searcher likely hedges on a centralized exchange. Their estimate is that 19 major searchers extracted $233.8M from 7,203,560 arbitrage trades between August 2023 and March 2025. They also show that the market is centralizing: three searchers capture about 73% of extracted value, and searchers with exclusive builder relationships keep lower margins because they share most revenue with builders. If correct, this gives the first comprehensive accounting of CEX-DEX arbitrage economics under Ethereum's proposer-builder separation.

What carries the argument

The load-bearing object is each searcher's optimal execution horizon, $t^*_j$, defined as the markout horizon in $[-1, +10]$ seconds around slot time (sampled every 0.5 seconds) at which the cross-trade median gross return peaks for that searcher. Gross return is markout revenue per unit of trade volume, computed from centralized-exchange mid-prices. Revenue and net PnL for every trade are evaluated at this fitted horizon: estimated extracted value equals markout revenue minus base fees, and PnL subtracts builder tips. The argument also relies on six heuristics that identify CEX-DEX arbitrage transactions from on-chain data, including an extension to multi-swap transactions that previous single-swap filters missed.

What would settle it

Directly observe the centralized-exchange leg. If exchange-side records or voluntary disclosures for a large searcher showed hedge times spread across many seconds (or fragment and limit-order hedging) rather than concentrated at the fitted $t^*_j$, the revenue and PnL estimates would be biased; the paper's upper-bound interpretation would fail if hedging on illiquid ALT tokens incurs substantial price impact not captured by markout mid-prices.

Watch

Extended reading notes

Core claim

The central claim is a quantitative statement about who earns what from CEX-DEX arbitrage on Ethereum: a total of $233.8M in extracted value was earned by 19 labeled searchers across 7,203,560 trades in 19 months, with the three largest searchers taking about 73% of cumulative revenue. The paper argues that searcher profitability is shaped by integration with block builders: exclusive searchers pay a larger share of revenue to one builder, often leaving per-trade margins in the 20-40% range or below, while neutral searchers keep 30-70%. It further claims that previous builder-profit estimates were too low for vertically integrated builders because they ignored the searcher-side profit, and that including searcher PnL shrinks the apparent subsidy burden by roughly 15% fewer blocks. The paper frames these estimates as consistent upper bounds, since actual centralized-exchange hedge execution is unobservable.

Load-bearing premise

The estimates stand or fall on the assumption that each searcher can be represented by one fixed hedge time—the horizon at which that searcher's own historical median gross return peaks—and that all revenue is realized at that moment.

Editorial extensions

If this is right

  • Total CEX-DEX extracted value of $233.8M is comparable to all atomic arbitrage value since The Merge, making this a first-order MEV category.
  • The searcher market is highly concentrated and increasingly so: the top three searchers account for about 73% of cumulative extracted value and roughly 90% by Q1 2025.
  • Exclusive searcher-builder arrangements transfer most arbitrage revenue to builders, so measuring searcher PnL is necessary to see true builder profitability and subsidy behavior.
  • Vertical integration and exclusive order flow reinforce each other: higher searcher flow predicts higher builder market share one day later, and builder share predicts further flow back.
  • Smaller searchers with higher per-trade margins are losing volume share over time, indicating rising entry barriers and economies of scale.

Reading between the lines

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

  • If actual hedge execution is fragmented, delayed, or executed via limit orders, the $233.8M figure is an upper bound rather than a point estimate; a natural extension is to cross-check against voluntary disclosures or exchange-side data from any major searcher.
  • The same 'peak-then-decay' markout logic could be applied to other non-atomic MEV such as cross-chain arbitrage, where the off-chain or cross-chain leg is also unobservable.
  • The mutual reinforcement between exclusive flow and builder market share suggests that protocol-level mitigations reducing block time may lower total extracted value without necessarily lowering concentration, since integrated players keep their structural advantage.
  • If solver markets and CEX-DEX searcher markets are dominated by the same entities, the paper's centralization result points to a consolidated MEV supply chain rather than a builder-only problem.
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

3 major / 5 minor

Summary. The paper studies CEX-DEX arbitrage on Ethereum over 19 months (August 2023 to March 2025). It refines existing heuristics to identify 8,723,233 candidate arbitrage transactions, labels 23 major searcher entities, and develops a markout-based framework that, for each searcher, selects an 'optimal execution horizon' t*_j as the horizon maximizing the median gross return over that searcher's own historical trades. Using this horizon, the paper estimates per-trade extracted value and PnL, and reports a total of 233.8M USD extracted by 19 searchers from 7,203,560 arbitrages, with the top three searchers capturing about 73% of cumulative extracted value. It then analyzes searcher-builder integration, profit margins, exclusive order-flow relationships, and corrects builder profitability for integrated searcher profits. The authors present their revenue estimates as upper-bound estimates in the body and discuss limitations, including unobservable hedge timing and hedge price impact.

Significance. If the quantitative estimates are reliable, this is the most comprehensive empirical measurement to date of CEX-DEX arbitrage economics on Ethereum and a valuable input to debates on MEV, PBS, and builder/searcher centralization. The paper's strengths include a long observation window, explicit and transparent detection heuristics, an open-source Dune query for the detection dataset, and several structural findings that are likely robust to the revenue-estimation details, such as the concentration of volume, the existence of exclusive searcher-builder flows, and the Kayle-Titan mutual reinforcement pattern. However, the headline revenue level, the top-three share, the profit margins in Table 1, and the builder-profit correction in Section 8 all depend on the per-searcher markout horizon t*_j, which is selected in-sample from the same trades being measured. Because the size of the resulting selection bias is not quantified, the central quantitative claim needs additional sensitivity and validation work before the paper's main number can be taken at face value.

major comments (3)
  1. [Section 4.1, Eq. (1); Section 6] The central estimate is computed at a per-searcher horizon t*_j = argmax over 23 markout horizons of the cross-trade median gross return on the same trades that are then valued at t*_j in Eq. (1). Even under the maintained assumption that each searcher hedges at a fixed true horizon, sampling noise makes the in-sample argmax exceed the revenue at the true horizon, so every headline quantity (the 233.8M total, the 73% top-three share, the Table 1 margins, and the Section 8 builder-profit correction) inherits an upward selection bias whose size is not bounded. The paper's 'sharp peaks / flat IQR' argument describes the shape of the median curve but does not quantify the selection effect. Please add a bootstrap or split-sample analysis (e.g., estimate t*_j on a rolling or random half of trades and evaluate on the held-out half), report the bias-corrected totals, and state in the abstract and Section 6 that the 233.8M figure is an upper-bound estimate, as Section 4.1 and Section 9 already acknowledge.
  2. [Sections 4.1 and 5.2] The markout revenue for ALT tokens is computed from Binance mid-prices without deducting hedge price impact, yet the paper itself shows that Pattern 2 searchers move prices sharply after their peak (the rapid closure of gross returns in Figure 3b). For low-liquidity tokens the observed post-peak decay is a direct measure of the cost the searcher pays to flatten inventory, and using the pre-hedge mid-price overstates revenue. This is acknowledged as a limitation but never quantified. Please report a sensitivity analysis that applies a haircut based on the observed post-peak decay (or on the peak-to-3s decline in Figure 5) and show how the ALT-token share of the 233.8M total changes.
  3. [Section 4.1 and Appendix H] The estimation pipeline drops 683,539 trades as 'inventory adjustment trades' because their markout revenue 'persistently fails to cover the base fees throughout the interval.' This filter is applied before t*_j is estimated and is itself based on the same markout pricing used to measure revenue, so it conditions the sample on trades that look profitable at some horizon and contributes an additional upward tilt to the estimates. The filter may be well motivated by the lower builder tips on those trades, but the paper does not report what happens to the headline total or to t*_j when these trades are retained with an alternative threshold. Please provide this robustness check.
minor comments (5)
  1. [Section 3.1] The phrase 'which roughly contribute to roughly 15%' contains a duplicated 'roughly' and should be edited.
  2. [Figure 2] The left panel's x-axis appears to begin at 2023-10 even though the text states the sample starts August 8, 2023; please align the axis or add a clarifying note.
  3. [Section 8] There are typos in 'upper-bound estimtate' and 'aggretated profit'; these should be corrected.
  4. [Section 4.1] The tie-breaking rule for t*_j (selecting the largest t when multiple horizons attain the same maximum) is stated after the definition; its effect on searchers with flat gross-return patterns is not discussed, even though the rule is relevant to the Pattern 3 exclusion.
  5. [Abstract and Section 6] The abstract and Section 6 present the 233.8M figure as a point estimate. Since the body repeatedly emphasizes that this is an upper-bound estimate, the abstract should carry the same qualifier for consistency.

Circularity Check

1 steps flagged · score 6.0 of 10

In-sample fitting of optimal execution horizons (Section 4.1) inflates the headline 233.8M USD estimate; structural centralization results remain independent.

  1. fitted input called prediction [Section 4.1, Eq. (1); headline total in Section 6; downstream Section 8]
    "For each searcher 𝑗 we define the optimal execution horizon 𝑡* 𝑗 as the markout horizon that maximizes the cross-trade median of gross return: 𝑡* 𝑗 = arg max 𝑡∈𝒯 Median 𝑖∈ℐ𝑗 {GR𝑖(𝑡)}, where 𝒯 = {−1.0, −0.5, 0, ..., 10.0} seconds ... Given 𝑡* 𝑗, the revenue (i.e., extractable value) and net profit and loss (PnL) for any trade 𝑖 ∈ ℐ 𝑗 are estimated as EV𝑖 = MR𝑖(𝑡* 𝑗) − base fees𝑖, and PnL𝑖 = EV𝑖 − builder tips𝑖."

    The unobservable hedge timing is not identified from external hedge data; it is chosen for each searcher as the markout horizon that maximizes the median gross return of the very same trades whose revenue is then measured. Eq. (1) values every trade at this in-sample argmax, so the headline total of 233.8M USD (Section 6), per-searcher revenues, PnL estimates, Table 1 margins, and the Section 8 integrated-builder profit correction are all computed at a horizon selected to make the estimated median return as large as possible among the 23 candidate horizons. This is an in-sample selection, not an independent estimate of realized hedge timing, and the paper provides no bootstrap or placebo quantification of the resulting upward bias.

full rationale

The only material circularity is the revenue-estimation step: the optimal execution horizon t*_j is fitted from each searcher's historical gross-return curve, and the same curve is then used to produce the EV and PnL numbers that support the headline 233.8M total. This is not a pure definitional identity, because t*_j is a timing parameter and the reported total is a sum rather than the fitted median, but the magnitude is nonetheless selected, not independently validated. The paper is transparent about some limitations, acknowledging uniform hedge timing and neglected hedge price impact, but it does not address the in-sample argmax selection bias. The other load-bearing components—transaction heuristics, searcher-builder volume shares, HHI concentration, and the exclusivity correlations—are based on observable on-chain data and do not reduce to the fitted horizon. The paper's self-citations (e.g., [51], [64], [67]) are used for background facts and prior estimates, not as the derivation's core, so they do not add to the circularity score. Because the central quantitative estimate is partly self-referential while most structural findings are independent, a score of 6 reflects partial circularity rather than full derivation-by-definition.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The main burden is not new entities but the unobservable CEX hedge. The paper replaces direct observation with a per-searcher fitted timing parameter, plus several market-structure assumptions about fee tiers, stablecoin parity, and liquidation of inventory.

free parameters (3)
  • Per-searcher optimal execution horizon t*_j = 0.5s to 2.0s (Table 7)
    Chosen as arg max of median gross return on the same trades used to compute EV and PnL; this is a fitted parameter central to all revenue estimates (Eq. 1).
  • Assumed CEX taker fee rate = 0.01725%
    Assumes all searchers use the highest Binance fee tier (Section 4.1 note 2); lowers costs and inflates revenue estimates if actual fees are higher.
  • Markout horizon window and step = [-1, 10] s, 0.5 s step
    Manual choice bounding all t*_j; extending the window would change pattern classification and could capture later hedges (Sections 4.1, 5.2).
assumptions (6)
  • domain assumption Searchers hedge on CEX only after their DEX trade is confirmed on-chain.
    Section 4: 'we assume that searchers first establish token inventory positions through on-chain DEX trades before subsequently hedging these accumulated positions on off-chain CEX venues.' Load-bearing for markout model.
  • ad hoc to paper Each searcher hedges at a single per-searcher optimal execution horizon t*_j inferred from the median gross return peak.
    Section 4.1: 'we assume each searcher initiates their hedge at their optimal execution horizon empirically identified from all their historical trades.' No direct observation of CEX leg.
  • domain assumption Binance USDT mid-price at the markout horizon is an executable USD price for the hedge, with no slippage or impact.
    Section 4.1 MR formula; especially fragile for ALT tokens where order books are thin.
  • domain assumption 1 USDT = 1 USD.
    Remark 1, Section 4.1.
  • domain assumption The heuristics (private, first-swap, non-atomic, Binance-listed tokens) isolate CEX-DEX arbitrage transactions and exclude other MEV or retail trades.
    Section 3, Heuristics 1-6; misclassification directly changes the dataset and totals.
  • domain assumption No off-chain rebates or Orderflow Auction refunds affect searcher PnL.
    Section 9 limitations: 'we do not account for potential off-chain rebates or Orderflow Auction refunds.'

how reviews work

0 comments
Cite this review

Pith. "Pith review of Measuring CEX-DEX Extracted Value and Searcher Profitability: The Darkest of the MEV Dark Forest." pith.science (2026). https://pith.science/paper/WRTWXON2

@misc{pith2026250713023,
  author       = {Pith},
  title        = {Pith review of: Measuring CEX-DEX Extracted Value and Searcher Profitability: The Darkest of the MEV Dark Forest},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WRTWXON2}},
  note         = {Machine review of arXiv:2507.13023}
}
read the original abstract

This paper provides a comprehensive empirical analysis of the economics and dynamics behind arbitrages between centralized and decentralized exchanges (CEX-DEX) on Ethereum. We refine heuristics to identify arbitrage transactions from on-chain data and introduce a robust empirical framework to estimate arbitrage revenue without knowing traders' actual behaviors on CEX. Leveraging an extensive dataset spanning 19 months from August 2023 to March 2025, we estimate a total of 233.8M USD extracted by 19 major CEX-DEX searchers from 7,203,560 identified CEX-DEX arbitrages. Our analysis reveals increasing centralization trends as three searchers captured three-quarters of both volume and extracted value. We also demonstrate that searchers' profitability is tied to their integration level with block builders and uncover exclusive searcher-builder relationships and their market impact. Finally, we correct the previously underestimated profitability of block builders who vertically integrate with a searcher. These insights illuminate the darkest corner of the MEV landscape and highlight the critical implications of CEX-DEX arbitrages for Ethereum's decentralization.

Figures

Figures reproduced from arXiv: 2507.13023 by the authors.

Figure 1
Figure 1. An example of a CEX-DEX arbitrage operation. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Daily count and weekly volume of detected CEX-DEX trades between August 2023 and [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Three empirically observed patterns of gross return distributions across all labeled [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: a) The relationship between median trade volume and gross returns for each searcher, with bubble size indicating searchers’ total estimated revenue and color encoding their execution pattern. b) Cumulative Distribution Function (CDF) of gross returns across all trades …
Figure 5
Figure 5. Figure 5: Correlation between the decline in median gross returns within 3 seconds after peak and [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Cumulative CEX-DEX arbitrage revenue (left panel) and daily share of revenue (right [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: The proportion of CEX-DEX arbitrage volume from each labeled searcher (x-axis, sorted [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: a) Bubble plot illustrating searchers’ median profit margin (x-axis) and median estimated PnL per trade (y-axis). The color shows the searcher’s integration level. The bubble size represents the searcher’s total PnL. Due to Graves having a negative total PnL, we show i…
Figure 9
Figure 9. Figure 9: a) Revenue (median) of exclusive searchers’ trades in their respective exclusive builders’ blocks vs. in other blocks. b) Profit margin (median) of exclusive searchers’ trades in their respective exclusive builders’ blocks vs. in other blocks. c) Percentage of total pa…
Figure 10
Figure 10. Figure 10: a) Titan market share and Kayle’s cumulative payments to Titan. The left y-axis shows the builder market share, and the right y-axis shows Kayle’ cumulative payments to Titan. b) rsync market share and Wintermute profit margin. The left y-axis shows the builder market…
Figure 11
Figure 11. Figure 11: Correlation analysis of Kayle’s daily volume share in Titan’s blocks (as a percentage of total daily CEX-DEX volume) and Titan’s daily market share. Using percentage metrics effectively controls for overall fluctuations in daily total CEX-DEX volume. opportunities. Bu…
Figure 12
Figure 12. Figure 12: The gross return distribution and pattern of the remaining searchers. [PITH_FULL_IMAGE:figures/full_fig_p032_12.png]
Figure 13
Figure 13. Figure 13: Daily DEX trade count and volume from atomic arbitrages, sandwich attacks, and [PITH_FULL_IMAGE:figures/full_fig_p033_13.png]
Figure 14
Figure 14. Figure 14: Weekly share of CEX-DEX volume by all the searchers, including Pattern 3 and other [PITH_FULL_IMAGE:figures/full_fig_p034_14.png]
Figure 15
Figure 15. Figure 15: HHI Index of weekly extracted value from CEX-DEX arbitrages by 19 labeled searchers. [PITH_FULL_IMAGE:figures/full_fig_p034_15.png]
Figure 16
Figure 16. Figure 16: Weekly share of CEX-DEX volume in Top 20 builders’ blocks and the HHI index. [PITH_FULL_IMAGE:figures/full_fig_p035_16.png]
Figure 17
Figure 17. Figure 17: a) beaverbuild market share, SCP median profit margin, and Senna’s cumulative payments to beaverbuild. The left y-axis shows the builder market share and SCP median profit margin, and the right y-axis shows Senna’s cumulative payments to beaverbuild. b) beaverbuild ma…

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Where Does MEV Really Come From? Revisiting CEXDEX Arbitrage on Ethereum

    cs.CR 2026-04 unverdicted novelty 7.0 of 10

    A new discrete-time AMM model with diffusive plus jump price processes shows CEX-DEX arbitrage requires volumes comparable to major liquidity pools and produces profits on the scale of total MEV.

  2. Geographical Centralization Resilience in Ethereum's Block-Building Paradigms

    cs.CR 2025-09 unverdicted novelty 6.0 of 10

    Ethereum's block-building paradigms create location-dependent incentives that drive geographical centralization of validators, quantified through mean-field analysis and latency-calibrated simulations.

Reference graph

Works this paper leans on

78 extracted references · 63 canonical work pages · cited by 2 Pith papers

  1. [1]

    URL: https://intel.arkm.com/

    Arkham. URL: https://intel.arkm.com/. 22

    work page

  2. [2]

    URL: https://coinmarketcap.com/

    Coinmarketcap. URL: https://coinmarketcap.com/

    work page

  3. [3]

    URL: https://dune.com/home

    Dune analytics. URL: https://dune.com/home

    work page

  4. [4]

    URL: https://etherscan.io/

    Etherscan.io. URL: https://etherscan.io/

    work page

  5. [5]

    URL: https://libmev.com/

    libmev. URL: https://libmev.com/

    work page

  6. [6]

    URL: https://zeromev.org/

    Zeromev. URL: https://zeromev.org/

    work page

  7. [7]

    Eip-7782: Reduce block latency

    Ben Adams and Dankrad Feist. Eip-7782: Reduce block latency. URL: https://eips.ether eum.org/EIPS/eip-7782

    work page

  8. [8]

    Optimal execution of portfolio transactions

    Robert Almgren and Neil Chriss. Optimal execution of portfolio transactions. Journal of Risk , 3(2):5–39, 2000

    work page 2000

Show all 78 references
  1. [9]

    The influence of cefi-defi arbitrage on order-flow auction bid profiles

    Data Always. The influence of cefi-defi arbitrage on order-flow auction bid profiles. https: //ethresear.ch/t/the-influence-of-cefi-defi-arbitrage-on-order-flow-auction-b id-profiles/17258. accessed: 2025-05-20

    2025

  2. [10]

    Mev bid data explorer

    Data Always. Mev bid data explorer. URL: https://dune.com/data_always/mev-bid-dat a-explorer?block_number_t72e94=22196045

  3. [11]

    Dex aggregator trades table

    Dune Analytics. Dex aggregator trades table. URL: https://docs.dune.com/data-catalog /evm/ethereum/curated-data/dex/dex-aggregator-trades

  4. [12]

    dex.trades table on ethereum

    Dune Analytics. dex.trades table on ethereum. URL: https://docs.dune.com/data-catal og/evm/ethereum/curated-data/dex/dex-trades

  5. [13]

    Barclay and Jerold B

    Michael J. Barclay and Jerold B. Warner. Stealth trading and volatility: Which trades move prices? Journal of Financial Economics , 34(3):281–305, 1993. URL: https://www.sciencedir ect.com/science/article/pii/0304405X9390029B, doi:10.1016/0304-405X(93)90029-B

    1993

  6. [14]

    Optimal control of execution costs

    Dimitris Bertsimas and Andrew W Lo. Optimal control of execution costs. Journal of financial markets, 1(1):1–50, 1998

    1998

  7. [15]

    Spot trading fee rate

    Binance. Spot trading fee rate. URL: https://www.binance.com/fee/trading

  8. [16]

    Mev blocker

    MEV Blocker. Mev blocker. https://mevblocker.io/. accessed: 2025-05-14

    2025

  9. [17]

    Mev capture and decentralization in execution tickets, 2024

    Jonah Burian, Davide Crapis, and Fahad Saleh. Mev capture and decentralization in execution tickets, 2024. URL: https://arxiv.org/abs/2408.11255, arXiv:2408.11255

    2024 arXiv

  10. [18]

    Becoming immutable: How ethereum is made, 2025

    Andrea Canidio and Vabuk Pahari. Becoming immutable: How ethereum is made, 2025. URL: https://arxiv.org/abs/2506.04940, arXiv:2506.04940

    2025 arXiv

  11. [19]

    Empirical analysis of cross domain cex-dex arbitrage on ethereum

    Colin Chan, Thomas Thiery, Julian Ma, and Barnab´ e Monnot. Empirical analysis of cross domain cex-dex arbitrage on ethereum. URL: https://ethresear.ch/t/empirical-analy sis-of-cross-domain-cex-dex-arbitrage-on-ethereum/17620

  12. [20]

    A tale of two arbitrages

    Eugene Chen, Alex Toberoff, Suraj Srinivasan, and Ankit Chiplunkar. A tale of two arbitrages. URL: https://frontier.tech/a-tale-of-two-arbitrages . 23

  13. [21]

    A new game in town

    Ankit Chiplunkar and Stephane Gosselin. A new game in town. URL: http://frontier.tec h/a-new-game-in-town

  14. [22]

    Statistical arbitrage on amms and block building on ethereum, 2024

    Mateusz Dominiak Christoph Rosenmayr. Statistical arbitrage on amms and block building on ethereum, 2024. URL: https://greenfield.xyz/2024/09/10/statistical-arbitrage-o n-amms-and-block-building-on-ethereum-part-1/

    2024

  15. [23]

    Flash boys 2.0: Frontrunning, transaction reordering, and consensus instability in decentralized exchanges

    Philip Daian, Steven Goldfeder, Tyler Kell, Yunqi Li, Xueyuan Zhao, Iddo Bentov, Lorenz Breidenbach, and Ari Juels. Flash boys 2.0: Frontrunning, transaction reordering, and consensus instability in decentralized exchanges. arXiv preprint arXiv:1904.05234 , 2019

    1904 arXiv

  16. [24]

    Committee-driven mev smoothing

    Francesco D’Amato. Committee-driven mev smoothing. URL: https://ethresear.ch/t/com mittee-driven-mev-smoothing/10408

  17. [25]

    Mev burn — a simple design

    Justin Drake. Mev burn — a simple design. URL: https://ethresear.ch/t/mev-burn-a-s imple-design/15590

  18. [26]

    Does a central clearing counterparty reduce counterparty risk? The Review of Asset Pricing Studies , 1(1):74–95, 2011

    Darrell Duffie and Haoxiang Zhu. Does a central clearing counterparty reduce counterparty risk? The Review of Asset Pricing Studies , 1(1):74–95, 2011

    2011

  19. [27]

    Buildernet

    Flashbots. Buildernet. URL: http://buildernet.org/docs

  20. [28]

    Dex trading bot table

    Flashbots. Dex trading bot table. URL: https://dune.com/queries/3375587

  21. [29]

    Illuminating the order flow landscape

    Flashbots. Illuminating the order flow landscape. https://writings.flashbots.net/illum inate-the-order-flow. accessed: 2025-05-23

    2025

  22. [30]

    League of legends dashboard

    Flashbots. League of legends dashboard. URL: https://dune.com/flashbots/lol

  23. [31]

    Mempool dumpster

    Flashbots. Mempool dumpster. accessed: 2025-05-08. URL: https://github.com/flashbots /mempool-dumpster

    2025

  24. [32]

    Mev-boost

    Flashbots. Mev-boost. https://docs.flashbots.net/flashbots-mev-boost/ . accessed: 2025-05-20

    2025

  25. [33]

    Mev-boost analytics - relay scan

    Flashbots. Mev-boost analytics - relay scan. https://www.relayscan.io/builder-profit? t=7d. accessed: 2025-05-14

    2025

  26. [34]

    Mev-share

    Flashbots. Mev-share. https://docs.flashbots.net/flashbots-protect/mev-share . accessed: 2025-05-14

    2025

  27. [35]

    Uniswap x fillers

    Flashbots. Uniswap x fillers. URL: https://dune.com/queries/4050099

  28. [36]

    Proof-of-stake (pos)

    Ethereum Foundation. Proof-of-stake (pos). https://ethereum.org/developers/docs/con sensus-mechanisms/pos, 2022. accessed: 2025-05-23

    2022

  29. [37]

    Censorship resistance in on-chain auctions, 2023

    Elijah Fox, Mallesh Pai, and Max Resnick. Censorship resistance in on-chain auctions, 2023. URL: https://arxiv.org/abs/2301.13321, arXiv:2301.13321

    2023 arXiv

  30. [38]

    Measuring arbitrage losses and profitability of amm liquidity,

    Robin Fritsch and Andrea Canidio. Measuring arbitrage losses and profitability of amm liquidity,

  31. [39]

    The centralizing effects of private order flow on proposer-builder separation

    Tivas Gupta, Mallesh M Pai, and Max Resnick. The centralizing effects of private order flow on proposer-builder separation. In 5th Conference on Advances in Financial Technologies , 2023

    2023

  32. [40]

    Ethereum’s proposer-builder separation: Promises and realities

    Lioba Heimbach, Lucianna Kiffer, Christof Ferreira Torres, and Roger Wattenhofer. Ethereum’s proposer-builder separation: Promises and realities. In 2023 ACM Internet Measurement Conference (IMC), Montreal, QC, Canada , October 2023

    2023

  33. [41]

    Non-Atomic Arbitrage in Decentralized Finance

    Lioba Heimbach, Vabuk Pahari, and Eric Schertenleib. Non-Atomic Arbitrage in Decentralized Finance . In 2024 IEEE Symposium on Security and Privacy (SP) , pages 3866–3884, Los Alamitos, CA, USA, May 2024. IEEE Computer Society. URL: https://doi.ieeecomputer society.org/10.1109...

    2024

  34. [42]

    Atomic mev table

    hildobby. Atomic mev table. URL: https://dune.com/queries/3493305

  35. [43]

    Mev sandwich trades

    hildobby. Mev sandwich trades. URL: https://dune.com/hildobby/sandwiches

  36. [44]

    Competition for retail order flow and market quality

    Edwin Hu and Dermot Murphy. Competition for retail order flow and market quality. Available at SSRN 4070056 , 2022

    2022

  37. [45]

    Korajczyk and Dermot Murphy

    Robert A. Korajczyk and Dermot Murphy. Performance of algorithmic trading orders: Size, speed, and volume. Review of Financial Studies , 32(7):2829–2873, 2019

    2019

  38. [46]

    Angstrom

    Sorella Labs. Angstrom. URL: https://x.com/angstromxyz/status/1889725902851854520

  39. [47]

    Moallemi, and Tim Roughgarden

    Jason Milionis, Ciamac C. Moallemi, and Tim Roughgarden. Automated market making and arbitrage profits in the presence of fees, 2023. URL: https://arxiv.org/abs/2305.14604 , arXiv:2305.14604

    2023 arXiv

  40. [48]

    Automated market making and loss-versus-rebalancing

    Jason Milionis, Ciamac C Moallemi, Tim Roughgarden, and Anthony Lee Zhang. Automated market making and loss-versus-rebalancing. arXiv preprint arXiv:2208.06046 , 2022

    2022 arXiv

  41. [49]

    Loss-versus-rebalancing under deterministic and generalized block-times, 2025

    Alex Nezlobin and Martin Tassy. Loss-versus-rebalancing under deterministic and generalized block-times, 2025. URL: https://arxiv.org/abs/2505.05113, arXiv:2505.05113

    2025 arXiv

  42. [50]

    Unity is strength: A formalization of cross-domain maximal extractable value, 2021

    Alexandre Obadia, Alejo Salles, Lakshman Sankar, Tarun Chitra, Vaibhav Chellani, and Philip Daian. Unity is strength: A formalization of cross-domain maximal extractable value, 2021. arXiv:2112.01472

    2021 arXiv

  43. [51]

    Who wins ethereum block building auctions and why? In 6th Conference on Advances in Financial Technologies , page 1, 2024

    Burak ¨Oz, Danning Sui, Thomas Thiery, and Florian Matthes. Who wins ethereum block building auctions and why? In 6th Conference on Advances in Financial Technologies , page 1, 2024

    2024

  44. [52]

    An empirical study of defi liquidations: Incentives, risks, and instabilities

    Kaihua Qin, Liyi Zhou, Pablo Gamito, Philipp Jovanovic, and Arthur Gervais. An empirical study of defi liquidations: Incentives, risks, and instabilities. In Proceedings of the 21st ACM Internet Measurement Conference, pages 336–350, 2021

    2021

  45. [53]

    Quantifying blockchain extractable value: How dark is the forest? In 2022 IEEE Symposium on Security and Privacy (SP) , pages 198–214

    Kaihua Qin, Liyi Zhou, and Arthur Gervais. Quantifying blockchain extractable value: How dark is the forest? In 2022 IEEE Symposium on Security and Privacy (SP) , pages 198–214. IEEE, 2022. 25

    2022

  46. [54]

    Ultra sound relay bid adjustment

    Ultra Sound Relay. Ultra sound relay bid adjustment. URL: https://github.com/ultraso undmoney/docs/blob/main/bid_adjustment.md

  47. [55]

    Andrei Shleifer and Robert W. Vishny. The limits of arbitrage. The Journal of Finance , 52(1):35–55, 1997

    1997

  48. [56]

    Cex-dex bots label list

    Danning Sui and Angela Lu. Cex-dex bots label list. URL: https://dune.com/queries/337 5615

  49. [57]

    Cex-dex transactions 202308-202503

    Danning Sui and Fei Wu. Cex-dex transactions 202308-202503. URL: https://dune.com/que ries/4931834

  50. [58]

    Available token price data on binance

    Tardis.dev. Available token price data on binance. URL: https://api.tardis.dev/v1/exch anges/binance

  51. [59]

    Binance spot historical data

    Tardis.dev. Binance spot historical data. URL: https://docs.tardis.dev/historical-dat a-details/binance

  52. [60]

    Builder dominance and searcher dependence

    Titan and Frontier Resarch. Builder dominance and searcher dependence. URL: https: //frontier.tech/builder-dominance-and-searcher-dependence

  53. [61]

    Blockchain censorship

    Anton Wahrst¨ atter, Jens Ernstberger, Aviv Yaish, Liyi Zhou, Kaihua Qin, Taro Tsuchiya, Se- bastian Steinhorst, Davor Svetinovic, Nicolas Christin, Mikolaj Barczentewicz, et al. Blockchain censorship. In Proceedings of the ACM Web Conference 2024 , pages 1632–1643, 2024

    2024

  54. [62]

    Time to bribe: Measuring block construction market

    Anton Wahrst¨ atter, Liyi Zhou, Kaihua Qin, Davor Svetinovic, and Arthur Gervais. Time to bribe: Measuring block construction market. arXiv preprint arXiv:2305.16468 , 2023

    2023 arXiv

  55. [63]

    mevboost.pics

    Toni Wahrst¨ atter. mevboost.pics. URL:https://mevboost.pics/

  56. [64]

    From competition to central- ization: The oligopoly in ethereum block building auctions

    Fei Wu, Thomas Thiery, Stefanos Leonardos, and Carmine Ventre. From competition to central- ization: The oligopoly in ethereum block building auctions. arXiv preprint arXiv:2412.18074 , 2024

    2024

  57. [65]

    Strategic bidding wars in on-chain auctions

    Fei Wu, Thomas Thiery, Stefanos Leonardos, and Carmine Ventre. Strategic bidding wars in on-chain auctions. In 2024 IEEE International Conference on Blockchain and Cryptocurrency (ICBC), pages 503–511, 2024. doi:10.1109/ICBC59979.2024.10634354

    2024

  58. [66]

    Searcher-builder relationship dashboard

    Winnie Xiao. Searcher-builder relationship dashboard. URL: https://www.searcherbuilder. pics/

  59. [67]

    Decentralization of ethereum’s builder market

    Sen Yang, Kartik Nayak, and Fan Zhang. Decentralization of ethereum’s builder market. In 2025 IEEE Symposium on Security and Privacy (SP) , page 1512–1530. IEEE, May 2025. URL: http://dx.doi.org/10.1109/SP61157.2025.00157, doi:10.1109/sp61157.2025.00157

    2025

  60. [68]

    Execution welfare across solver-based dexes,

    Yuki Yuminaga, Dex Chen, and Danning Sui. Execution welfare across solver-based dexes,

  61. [69]

    Rediswap: Mev redistribution mechanism for cfmms, 2024

    Mengqian Zhang, Sen Yang, and Fan Zhang. Rediswap: Mev redistribution mechanism for cfmms, 2024. URL: https://arxiv.org/abs/2410.18434, arXiv:2410.18434. 26

    2024 arXiv

  62. [70]

    High-frequency trading on decentralized on-chain exchanges

    Liyi Zhou, Kaihua Qin, Christof Ferreira Torres, Duc V Le, and Arthur Gervais. High-frequency trading on decentralized on-chain exchanges. In 2021 IEEE Symposium on Security and Privacy (SP), pages 428–445. IEEE, 2021

    2021

  63. [71]

    Cross-chain arbitrage: The next frontier of mev in decentralized finance, 2025

    Burak ¨Oz, Christof Ferreira Torres, Christoph Schlegel, Bruno Mazorra, Jonas Gebele, Filip Rezabek, and Florian Matthes. Cross-chain arbitrage: The next frontier of mev in decentralized finance, 2025. URL: https://arxiv.org/abs/2501.17335, arXiv:2501.17335. 27 A Data Collecti...

    2025 arXiv

  64. [74]

    Consequently, our analysis excludes arbitrage transactions involving tokens listed exclusively on other CEXes

    We only include transactions involving tokens listed on Binance. Consequently, our analysis excludes arbitrage transactions involving tokens listed exclusively on other CEXes. This choice likely results in an underestimation of the total revenue and profitability (PnL) of some...

  65. [75]

    This heuristic helps filter out retail or unrelated trades but may also exclude valid arbitrage trades that occur after other swaps

    We only consider transactions in which at least one swap executed is the first trade in its direction within the corresponding DEX pool in that block. This heuristic helps filter out retail or unrelated trades but may also exclude valid arbitrage trades that occur after other ...

  66. [76]

    Finally, cross-chain arbitrage transactions (i.e., arbitrages executed between decentralized exchanges on different blockchains) might occasionally be misclassified as CEX-DEX arbitrages. Although we have verified that known cross-chain arbitrageur addresses identified in prio...

  67. [77]

    The transaction consumes less than 400,000 gas

    “The transaction consumes less than 400,000 gas. ” We expanded our analysis to include transactions involving multiple token swaps and ERC-20 token transfers, which inherently consume more gas than single-swap transactions. Consequently, the previous gas consumption limit of 4...

  68. [78]

    The transaction includes a coinbase transfer or a priority fee of at least 1 GWei

    “The transaction includes a coinbase transfer or a priority fee of at least 1 GWei. ” Theoretically, integrated searchers do not need to send explicit tips for inclusion in the block built by their integrated builders. Therefore, this heuristic is removed to capture trades by ...

    2023

  69. [2024]

    URL: https://arxiv.org/abs/2404.05803, arXiv:2404.05803. 24

    arXiv

  70. [2025]

    URL: https://arxiv.org/abs/2503.00738, arXiv:2503.00738

    arXiv

Pith tools

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