Three Essays on Financial Markets as Complex Dynamical Systems

This dissertation consists of three self-contained chapters in the domain of financial markets and complex systems, all of them driven by the choice to model and describe modern financial markets from a network and data science perspectives. In the first chapter of this manuscript I try to uncover stable community structures in the stock market, redefining the classical mean-variance portfolio optimization framework in terms of communities to which stocks belong. This will lead to the calculation of an optimal intra-community equally weighted portfolio, whose reliability will be addressed in terms of discrepancies between the predicted and realized variances. The second chapter models the stock market relying again on network representations but from an econometric modeling perspective. The objective here is to show how the topological properties of econometric inferred networks can be exploited to map the explanatory power of entities as sources of cross-country volatility spillovers. Finally, the third chapter of this manuscript will move towards high-frequency financial markets. The objective will be to identify, through data-driven identification procedures for vector error correction models (VECM), the leaders and the followers of the price formation process for assets contemporaneously traded on multiple exchanges.