Academic year 2021–22
How to Build an Arbitrage Bot
A co-authored research and benchmark study of the architecture behind a decentralized-exchange arbitrage searcher on Ethereum.
- Context
- Research project at the Polytechnic University of Milan, co-authored with Elia Maggioni and Lucas Manini.
- Role
- Co-author
- Contribution
- Co-authored a model for cyclic arbitrage across decentralized exchanges and a two-stage architecture: startup-time market-graph construction followed by block-driven state synchronization and arbitrage-path evaluation.
- Outcome
- Built a benchmark implementation to examine graph-construction space and runtime costs, including synchronization and arbitrage checks. The report documents the architecture, complexity analysis, and benchmark results.
This report studies automated arbitrage across Ethereum-based decentralized exchanges. It models trade cycles across liquidity pools and describes how to search for opportunities as new blocks arrive.
The proposed system separates startup-time work, which builds the market graph and candidate paths, from a runtime module that syncs pool state and evaluates those paths. A Go benchmark implementation is used to analyze graph-construction space requirements and runtime costs. The report focuses on system architecture and performance evaluation.