ETD

• complex systems
• eMID
• entropy
• interbank market
• maximum likelihood
• statistical mechanics
• unsecured interbank market

Over recent decades, network theory has become a key method to study complex systems, including financial ones. This Thesis applies the statistical mechanics framework to unsecured interbank credit networks, showing that a simple canonical ensemble reproducing the observed core–periphery asymmetry can replicate real market organization with fewer parameters than classical null models. The unsecured interbank market, exemplified by Italian e-MID data, reveals sparse, disassortative, heavy-tailed structures motivating such modeling.
A mixed constraint maximum entropy model is introduced, combining a global field with core-specific fitnesses to ensure convexity, interpretability and scalability. Analytical results yield closed form probabilities and phase regimes, validated through Monte Carlo simulations and empirical data. Compared to SBM benchmarks, the model better predicts higher order patterns and traces structural shifts during 2009–2014. Overall, the Thesis bridges Physics and Finance, offering a minimal yet testable framework for systemic risk monitoring.