In this thesis, we investigate the financial stability of blockchain-based markets and the stability of blockchain protocols themselves through empirical, data-driven analysis. It comprises three studies addressing stability at the market, protocol, a...
In this thesis, we investigate the financial stability of blockchain-based markets and the stability of blockchain protocols themselves through empirical, data-driven analysis. It comprises three studies addressing stability at the market, protocol, and mechanism levels.
First, we examine market-level stability by analyzing price co-movements among DeFi tokens, a key proxy for systemic risk and the effectiveness of portfolio diversification. Using correlation-based techniques, we identify persistent synchronization in token prices, which intensifies during bear markets and weakens in bull markets. The results highlight that DeFi tokens traded on decentralized exchanges exhibit stronger co-movement than cryptocurrencies on centralized exchanges, suggesting systemic coupling within the DeFi ecosystem.
Second, we explore protocol-level vulnerabilities by addressing Maximal Extractable Value (MEV), a key source of instability and unfairness in blockchain networks. We propose ArbiNet, a novel graph neural network (GNN)-based model that detects MEV activities without relying on pre-registered Application Binary Interfaces (ABIs). Trained on an extensive, newly constructed MEV dataset, ArbiNet achieves efficient and accurate detection, offering insights into the behavioral dynamics that undermine blockchain security and fairness.
Finally, we empirically evaluate the impact of Ethereum Improvement Proposal EIP-4844, which introduces blob-carrying transactions to enhance data availability. Using on-chain data, we assess its effects on consensus security, rollup transaction dynamics, and blob gas fee markets. Our findings show that EIP-4844 effectively reduces rollup transaction costs and increases network utilization, though minor trade-offs appear in fork rates and transaction inclusion latency.
Collectively, these studies provide empirical insights into how blockchain systems interact with stability at multiple levels—market, protocol, and design—providing a critical, multi-level framework for assessing the stability and maturity of future decentralized financial systems.