An Empirical Analysis of DEX Arbitrage on Ethereum

The rise in popularity and liquidity on Ethereum over the past several years has attracted a large number of traders and researchers to the platform. When combined with the decentralized nature of the Ethereum blockchain, arbitrage opportunities quickly become intuitive, and their existence should be evident in the DEX ecosystem. The majority of previous research on this topic agrees, with claims of tens of millions of dollars of on-chain arbitrage profits living on Ethereum. To test these hypotheses, we developed a novel Multi-Cycle Moore-Bellman-Ford algorithm capable of detecting multiple unique arbitrage cycles per block within Ethereum’s 12-second time constraint. Our methodology includes rigorous data cleaning, fee calculation, graph construction, cycle detection using MC-MBF, profitability optimization via the bisection method, and transaction simulation. Our results demonstrate that after analyzing more than 29 DEXs, 7,026 tokens, and 25,811 liquidity pools, most arbitrage opportunities identified in theory become unprofitable after accounting for gas fees, constant and dynamic liquidity pool fee structures, slippage, and on-chain time constraints.