Paper: arXiv 2507.13023

Authors: Fei Wu, Danning Sui, Thomas Thiery, Mallesh Pai

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.

Complexity vs Empirical Score

  • Math Complexity: 3.5/10
  • Empirical Rigor: 6.0/10 (capped from 8.0: no out-of-sample test; see penalty flags)
  • Quadrant: Street Traders — empirical and applied, lower math complexity (effective empirical rigor capped from 8.0)

Why this score: The paper relies primarily on large-scale empirical data analysis, heuristics, and statistical estimation, with minimal heavy theoretical derivations or formulas. It demonstrates high empirical rigor through extensive data collection (19 months), detailed methodology for transaction identification, and concrete profit/loss calculations.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Quantify CEX-DEX arb economics & searcher profitability"] --> B["Methodology: Empirical on-chain analysis"]
  B --> C["Data: 19 months on Ethereum (Aug 2023 - Mar 2025)"]
  C --> D["Identification: Refined heuristics for CEX-DEX arb txs"]
  D --> E["Computation: Estimate revenue without CEX trade data"]
  E --> F["Analysis: Searcher concentration & builder relationships"]
  F --> G["Key Findings:<br>• $233.8M extracted by 19 searchers<br>• Top 3 searchers capture 75% volume<br>• Vertical integration boosts builder profitability<br>• Exclusive searcher-builder relationships exist"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper refines heuristics for identifying CEX‑DEX arbitrage transactions and introduces an empirical framework to estimate arbitrage revenue without observing the CEX leg.

Source sentence
We refine existing heuristics for identifying CEX-DEX arbitrage transactions [41, 51], signifi‑cantly expanding coverage and accuracy.

Assumes 6

  • Searchers first establish token inventory on‑chain before hedging on off‑chain CEX venues. execution
    Source sentence
    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.
  • Searchers hedge both directions on CEX within a very short window after confirming on‑chain trade execution. execution
    Source sentence
    In addition, we assume that searchers execute the hedge for both directions on CEX within a short timeframe or even instantaneously after confirming successful on-chain trade.
  • 1 USDT is equal to 1 USD for pricing purposes. model
    Source sentence
    We assume 1 USDT = 1 USD, as most tokens are priced in USDT on Binance.
  • Uniform hedge execution timing across all trades for each searcher. model
    Source sentence
    Our revenue estimation method—based on empirically derived optimal execution horizons across historical trades for each searcher—introduces approximation errors by assuming uniform hedge execution timing across all trades for each searcher.
Show 2 more
  • Price impact of hedging is neglected, especially for low‑liquidity tokens. model
    Source sentence
    neglecting the price impact of hedging, particularly relevant for low‑liquidity tokens.
  • Off‑chain rebates or Orderflow Auction refunds are not accounted for. other
    Source sentence
    Finally, we do not account for potential off-chain rebates or Orderflow Auction refunds, which might underestimate actual searcher PnL and overstate builder profits.

Data

  • CEX‑DEX arbitrage transaction dataset crypto-onchain unknown
    August 2023 to March 2025 · 19 major CEX‑DEX searchers, 7,203,560 identified arbitrages
    Source sentence
    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.
  • Binance USDT‑pair historical quote data · Binance via Tardis.dev market-prices public
    All tokens listed on Binance
    Source sentence
    by leveraging the superior liquidity of USDT pairs on Binance. To derive accurate USD prices for each token
  • MEV‑Boost auction data crypto-onchain unknown
    Source sentence
    Integrating DEX trades data, CEX pricing, and MEV‑Boost data over 19 months from August 2023 to March 2025…

Recognised sources: Binance Etherscan Tardis.dev CoinMarketCap

Costs

Unclear The classifier was not confident.

Finds 4

  • Searchers extracted an estimated total of 233.8 M USD from CEX‑DEX arbitrages over the sample period. performance
    Source sentence
    we estimate a total of 233.8M USD extracted by 19 major CEX‑DEX searchers from 7,203,560 identified CEX‑DEX arbitrages.
  • Three searchers captured roughly three‑quarters of both volume and extracted value, indicating increasing centralization. statistical
    Source sentence
    Our analysis reveals increasing centralization trends as three searchers captured three‑quarters of both volume and extracted value.
  • Pattern 1 searchers trade large (>10,000 USD) major‑token volumes with single‑digit bps gross returns; Pattern 2 searchers trade smaller (<5,000 USD) ALT tokens with higher (20‑120 bps) returns. performance
    Source sentence
    resulting in markedly higher gross returns ranging from 20 to 120 bps per trade.
  • Searchers trading major tokens experience slower and smaller declines in gross returns after peak, reflecting lower price impact. statistical
    Source sentence
    We find a strong negative correlation, confirming that searchers predominantly trading major tokens experience slower and smaller gross return declines, reflective of reduced price impact.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 27 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 4 papers in the archive (4 by independent authors): Arbitrage with bounded Liquidity (discusses its findings); Impact of Volatility on Time-Based Transaction Ordering … (uses its method or data)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (96)

  1. Mempool Dumpster — link
  2. Mempool Hygrometer — link
  3. $\alpha$-rank: Multi-agent evaluation by evolution (2019)
  4. Strategic Bidding Wars in On-chain Auctions (2024) — link
  5. Time to Bribe: Measuring Block Construction Market (2023) — link
  6. Time is Money: Strategic Timing Games in Proof-of-Stake Protocols (2023) — link
  7. Game-theoretic Model for MEV-boost Auctions (MMA)
  8. Empirical analysis of Builders’ Behavioral Profiles (BBPs)
and 88 more
  1. Bid cancellations considered harmful
  2. Order flow, auctions and centralisation I - a warning
  3. Order flow, auctions and centralisation II - order flow auctions
  4. Illuminating The Order Flow Landscape
  5. The Centralizing Effects of Private Order Flow on Proposer-Builder Separation (2023)
  6. The cost of artificial latency in the PBS context — → 2312.09654
  7. Timing Games: Implications and Possible Mitigations
  8. Structural Advantages for Integrated Builders in MEV-Boost (2023) — link
  9. Flash boys 2.0: Frontrunning, transaction reordering, and consensus instability in decentralized exchanges (2019) — link
  10. Unity is Strength: A Formalization of Cross-Domain Maximal Extractable Value (2021) — link
  11. Quantifying blockchain extractable value: How dark is the forest? (2022)
  12. MEV-Boost
  13. The Influence of CeFi-DeFi Arbitrage on Order-Flow Auction Bid Profiles
  14. Time Moves Faster When There is Nothing You Anticipate: The Role of Time in MEV Rewards (2023) — link
  15. Latency is Money: Timing Games
  16. Honest Validator - Ethereum Consensus Specifications
  17. Decentralization of Ethereum’s Builder Market (2025) — link
  18. MEV Blocker
  19. MEV-Share
  20. The Future of MEV is SUAVE
  21. The MEVM, SUAVE Centauri, and Beyond
  22. MEV-Boost Analytics - Relay Scan
  23. Who Wins Ethereum Block Building Auctions and Why? (2024)
  24. Non-Atomic Arbitrage in Decentralized Finance — → 2401.01622
  25. Atomic MEV table — link
  26. DEX trading bot table — link
  27. dex.trades table on Ethereum — link
  28. Usage of Markout to Calculate LP Profitability in Uniswap V3
  29. Usage of Markout to Calculate LP Profitability in Uniswap V3
  30. Spot Trading Fee Rate
  31. Binance Spot historical data
  32. Cross-Chain Arbitrage: The Next Frontier of MEV in Decentralized Finance (2025) — link
  33. CEX-DEX transaction dataset Aug 2023 to Dec 2024
  34. DEX pool autocorrelation Aug 2023 to Dec 2024
  35. Zeromev
  36. Etherscan.io
  37. CoinMarketCap
  38. Empirical Analysis of Cross Domain CEX-DEX Arbitrage on Ethereum
  39. Dune Analytics
  40. High-frequency trading on decentralized on-chain exchanges (2021)
  41. CEX-DEX bots label list
  42. A new game in town
  43. A Tale of Two Arbitrages
  44. Aestus Timing Games as a Services
  45. Arkham
  46. Searcher-Builder Relationship Dashboard
  47. League of Legends Dashboard
  48. LVR.wtf
  49. mevboost.pics
  50. Ultra Sound Relay Bid Adjustment
  51. Builder Dominance and Searcher Dependence
  52. MEV Bid Data Explorer
  53. Available Token price data on Binance
  54. Proof-of-Stake (PoS) (2022)
  55. An empirical study of defi liquidations: Incentives, risks, and instabilities (2021)
  56. Automated market making and loss-versus-rebalancing (2022) — link
  57. Automated Market Making and Arbitrage Profits in the Presence of Fees — → 2305.14604
  58. From Competition to Centralization: The Oligopoly in Ethereum Block Building Auctions (2024) — link
  59. Ethereum’s Proposer-Builder Separation: Promises and Realities (2023)
  60. Dex Aggregator Trades table
  61. Uniswap X Fillers
  62. Atomic Arbitrages
  63. MEV Sandwich Trades
  64. Execution Welfare Across Solver-based DEXes (2025) — link
  65. BuilderNet
  66. Committee-driven MEV smoothing
  67. MEV Burn — a simple design
  68. Angstrom
  69. Blockchain censorship (2024)
  70. Censorship Resistance in On-Chain Auctions (2023) — link
  71. libMEV
  72. Stealth trading and volatility: Which trades move prices? (1993) — link
  73. Optimal Execution of Portfolio Transactions (2000)
  74. Performance of Algorithmic Trading Orders: Size, Speed, and Volume (2019)
  75. The implementation shortfall: Paper versus reality (1988)
  76. Optimal control of execution costs (1998)
  77. Loss-Versus-Rebalancing under Deterministic and Generalized block-times — → 2505.05113
  78. MEV Capture and Decentralization in Execution Tickets — → 2408.11255
  79. CEX-DEX Transactions 202308-202503
  80. Becoming Immutable: How Ethereum is Made (2025) — link
  81. RediSwap: MEV Redistribution Mechanism for CFMMs (2024) — link
  82. Quantifying Arbitrage in Automated Market Makers: An Empirical Study of Ethereum ZK Rollups (2024)
  83. Statistical arbitrage on AMMs and block building on Ethereum (2024) — link
  84. Measuring Arbitrage Losses and Profitability of AMM Liquidity — → 2404.05803
  85. The Limits of Arbitrage (1997)
  86. Does a central clearing counterparty reduce counterparty risk? (2011)
  87. Competition for retail order flow and market quality (2022)
  88. EIP-7782: Reduce Block Latency

6 of 96 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.