A0371
Title: Exploring the impact of non-linear dependencies in stock market returns regime transitions
Authors: Marina Dolfin - King's College London (United Kingdom) [presenting]
Jose De Leon Miranda - KCL (United Kingdom)
Abstract: Stock markets exemplify complex systems characterized by non-linear dependencies among actors at different levels of observation: at a micro level, concerning individual stocks, and at a macro level, when multiple markets interact. Importantly, regime transitions in these complex systems may endogenously arise due to the non-linear interactions, often shaped by the inherent heterogeneity among the market actors, resulting in asymmetrical correlations. Based on these considerations, we explore the impact of non-linear dependencies in the dynamics of market returns by analysing their time-dependent cross-correlations. We employ and compare several approaches, including Detrended Cross-Correlation Analysis (DCCA), spatial correlations, autocorrelations (to detect bifurcations) and minimum spanning trees. The principal objective of our research is to discern patterns that emerge during regime phase transitions and to identify potential early warning signals preceding regime changes, including market crashes. Additionally, we propose investigating the network system's controllability as a future research perspective. The data under analysis comprises returns from stock indices across different geographical regions.
