Abstract of doctoral thesis - Pavlos Zitis

Machine Learning and Complex Systems Time Series Analysis in Quantitative Finance and Risk Management.

The present PhD dissertation refers to the analysis of the financial markets using methods of complexity theory and machine learning techniques.

The global financial system is constantly evolving. This is due to various factors such as technological progress, new investment products, new regulations and evolving consumer preferences. Hence, the analysis of the financial markets with traditional methods could be enhanced by new alternative methods of analysis.

In recent years, a new area of research related to complex systems research called "Econophysics" has been developed. Econophysics is an interdisciplinary field which applies the methods of statistical physics, nonlinear dynamics, and network theory to macro-micro/economic modeling, to financial market analysis and social problems. 

On the other hand, at the same period with the development of econophysics, a rapid advance in computer science started. This progress led to the growth of machine learning methods. With the advent of machine learning, it has become possible to develop new algorithms and strategies for data analytics of complex systems. Hence, exploring how machine learning works for issues involving complex systems such as the global financial system is a subject of significant research interest.

The main research objective of this PhD candidacy is the development of an alternative systemic risk measure using methods of complexity theory and machine learning techniques.

The present study is expected to significantly contribute to the creation of an automated early warning system for extreme events occurring in the financial markets. The development of such a system could be implemented by large financial institutions.