A Survey on Graph-Based Analysis of the DeFi Ecosystem

Abstract

Decentralized Finance (DeFi) has rapidly emerged as one of the most transformative applications of blockchain technology, enabling permissionless financial services including lending, borrowing, trading, and yield optimization through smart contract-based protocols. The complex interactions among users, protocols, and assets in the DeFi ecosystem naturally form intricate graph structures whose topology encodes critical information about system behavior, risk propagation, and economic vulnerabilities. This paper presents a comprehensive survey of graph-based analytical methods applied to the DeFi ecosystem. We systematically review transaction graphs, protocol dependency graphs, liquidity flow networks, and user interaction graphs, examining how graph-theoretic techniques-including community detection, centrality analysis, temporal graph mining, and anomaly detection-have been applied to understand DeFi dynamics. We further analyze how graph methods reveal systemic risks such as composability-induced contagion, flash loan attack patterns, and liquidity concentration. We propose a unified DeFi Graph Analysis Taxonomy (DGAT) that classifies existing work across graph construction methodologies, analytical objectives, and application domains. Comparative evaluation of current approaches identifies key research gaps and outlines directions for future investigation, particularly in cross-protocol risk modeling and real-time graph analytics for DeFi monitoring.