MEV-Boost Auction Bids Dataset

<p><strong>This dataset contains the data used in the paper <em>"Decentralization of Ethereum's Builder Market"</em> (IEEE S&P 2025). It includes the following components:</strong></p>

<hr />

<h3>1. Partial Bids Dataset</h3>

<p>This dataset contains partial bids collected from 13 MEV-Boost relays, covering the period from <strong>September 15, 2022, to October 31, 2024</strong>. A <strong>partial bid</strong> refers to a bid submitted by a builder that includes metadata—such as the block hash and bid value—but not the full block contents. In an MEV-Boost auction, relays aggregate bids from multiple builders and forward the highest bid to the proposer for inclusion.</p>

<p>We query each relay daily and store all received partial bids. Each record includes the block hash and the associated bid value submitted by a builder. The dataset is organized by day, allowing for temporal analysis of builder behavior and MEV-Boost auction dynamics.</p>

<hr />

<h3>2. Private Transactions Dataset</h3>

<p>This dataset tracks <strong>private transactions</strong>—transactions that are confirmed on-chain but do not appear in the public mempool. These transactions typically bypass public propagation and are submitted directly to block builders through private channels. The dataset is <strong>organized by month</strong>, with each file containing all detected private transactions for that period. It currently covers the period from <strong>September 15, 2022, to October 31, 2024</strong>.</p>

<p>To identify private transactions, we use a combination of full nodes and API subscriptions to QuickNode and Infura to collect pending transactions as they enter the mempool. We then compare this data to the on-chain transactions to identify private transactions. Our data collection process is described in more detail at <a href="https://mempool.guru">https://mempool.guru</a>.</p>