Optimization Methods for Generalized Covariance Estimator with an Application to Cryptocurrency

This research project aims to enhance estimators for nonlinear dynamic models with non-Gaussian errors. Objectives include improving the semi-parametric GCov estimator by addressing numerical challenges and proposing an optimized algorithm. Adjustments to the objective function will handle sparsity and singularity issues in big data applications. The project will apply the GCov estimator to analyze the joint modelling of cryptocurrency rates and financial assets, providing insights for risk mitigation strategies. Mixed causal-noncausal models, innovative specification tests, and the Simulated Annealing algorithm will be utilized. Furthermore, the project aims to develop an R package for the GCov estimator and associated tests, facilitating the application and validation of the proposed methodology. Policymakers, investors, and risk managers will benefit from the research, aiding decision-making and understanding the interconnections between cryptocurrencies and financial markets.

Faculty Supervisor:

Joann Jasiak

Student:

Partner:

Maastricht University

Discipline:

Sociology

Sector:

Education

University:

York University

Program: