A Multiplex Network Hawkes Model for Systemic Risk Measurement

Battle Count: 127

The paper is primarily focused on systemic risk measurement and financial network topology rather than direct trading strategies. However, the identification of systemically important institutions, contagion channel decomposition, and network-based risk transmission metrics are relevant for risk management in trading portfolios, counterparty risk assessment, and understanding market-wide stress propagation. The CDS-based event framework could inform credit-risk-aware trading strategies.

High complexity due to: (1) multiplex Hawkes process with L layers requiring joint inference over adjacency matrix, weight matrices, interaction kernels, and background rates; (2) adaptive Metropolis-Hastings with ASM/ASWAM for gamma regression parameters; (3) latent parent allocation with categorical sampling; (4) label-switching post-processing via ECR algorithm; (5) numerical stabilization for sparse networks with zero-weight edges; (6) 20,500 MCMC samples needed; (7) careful prior specification and sensitivity analysis required. The full algorithm spans multiple appendix sections with detailed derivations.

The paper provides detailed algorithm specifications (Algorithm 1), full prior specifications with default hyperparameter values, simulation scenarios with exact parameters (Table 2), and comprehensive appendix material including MCMC sampler details, adaptive MH algorithms, ECR label-switching procedures, and generative model description. However, no code repository is explicitly linked.