Hi everyone! We’re MEV-X, a research-driven team that’s been studying MEV mechanics on EVM-compatible chains for a while. Some of you may have seen our earlier work on atomic arbitrage classification on Polygon and on RL-based bidding strategies in sealed-bid auctions.
About six months ago we started working on a problem that felt important but underexplored: which prior transactions created the conditions for profitable arbitrage, and what was each one’s marginal contribution to the extracted value.
We submitted two grant proposals to FRP around this, both of which were declined, with reviewers ultimately concluding that combinatorial explosion in real blocks would make the approach infeasible. We believed the problem was tractable and pursued it independently, and we’re happy to share what came out of it!
To address this, we implement four distinct methodological approaches to the MEV attribution problem, each offering different trade-offs between analytical rigor, computational efficiency, and data requirements:
- Simulation-based: isolates the causal impact of specific transactions on arbitrage profitability via counterfactual execution
- Coefficient-based: analyzes the mathematical impact of transactions on pool pricing coefficients without requiring full state simulation
- Shapley-based: applies cooperative game theory to fairly attribute profit when multiple transactions may have contributed
- Bot-data-driven: leverages the behavior of production MEV searcher bots as a proxy for ground truth attribution
We evaluated our methods on 1,050,000 blocks on Polygon (March 2–29, 2026), covering 360,026 atomic arbitrage transactions with a total extracted MEV volume of $334,799.
Our key findings:
- The single-source hypothesis holds for 96.7% of cases: one transaction is responsible for the vast majority of the created opportunity, with the next largest contributor typically accounting for less than 5%
- Simulation-based attribution provides the best practical balance: 91.7% accuracy at 12.3ms per event with near-complete coverage (99.1%)
- MEV creation is highly concentrated: the top 5% of protocols generate 73% of opportunities, with Uniswap V3 participating in 58% of opportunity-creating transactions
The full paper is available on arXiv: The Origins of MEV: Systematic Attribution of Arbitrage Opportunity Creation at Scale
The current work covers atomic arbitrage on Polygon, where we validated the methodology and demonstrated its feasibility at scale. The natural next step is extending the framework to Ethereum mainnet.
Happy to discuss methodology (especially 
), findings, or where this goes next!