This repository contains the code for the paper "Structural Estimation of MEV-Boost Auctions" by Tivas Gupta, Mallesh M. Pai, Max Resnick, and Xun Tang.
This project implements a structural econometric model to estimate the distribution of builder values in MEV-Boost auctions on Ethereum. The model uses a two-stage non-linear least squares procedure to decompose builder values into common and idiosyncratic components, separately identifying type-specific factors from builder-specific private information.
pip install -r requirements.txtpython analysis.pyThe script performs two-stage estimation for integrated and neutral builders, calculates R-squared statistics, and runs bootstrap resampling (500 iterations) to compute 95% confidence intervals.
The dataset (aggregated_bids_with_volatility_and_mempool.csv) is too large for GitHub. The data contains approximately 95,000 MEV-Boost auctions from selected days between October and December 2024, including bid data from major builders, CEX price movements, base fees, and mempool statistics.
Data is available from the authors upon request.
If you use this code, please cite:
@article{gupta2025structural,
title={Structural Estimation of MEV-Boost Auctions},
author={Gupta, Tivas and Pai, Mallesh M. and Resnick, Max and Tang, Xun},
journal={Working Paper},
year={2025}
}