Price Dynamics of Automated Market Makers: Simulation-based Approach

The aim of this thesis is to analyze the price dynamics implied by the Automated Market Makers used by Decentralized Exchanges of DeFi and to verify the presence of some behavioral patterns with a simulation-based approach. Returns from 10 representa- tive token pairs were collected over a 15-day period and their properties were compared against traditional stylized facts. A simulation that reproduces the observed price pro- cess was then developed, mimicking the realized swap orders and utilizing the constant product pricing equation of AMMs, incorporating two additional features that implement periods of hype and herding behavior. Analysis of the empirical data revealed that AMM token returns follow the stylized facts most of the time, with their distributional properties, autocorrelation patterns, and volatility clustering. No consistent trend in the leverage effect was found among tokens. The simulation then confirmed that a basic AMM model is sufficient in producing prices with similar returns, showing that this method of transaction settlement is robust and generates the expected price dynamics. The two behavioral mechanics added further increased the similarity between real and simulated return characteristics, indicating that the effects may also influence the actual price formation process.