State-of-the-Art Measurement and 'Early Warning' of Critical Transitions in Financial Markets

One of the characteristic features of complex systems is the existence of critical transitions ("tipping points") that separate relatively stable regions of system behaviour. Such transitions have been observed in a variety of systems, from phase transitions between different states of matter, such as ice and liquid water, to changes in ecosystems, international economy and global climate. In this interdisciplinary project, we will study financial markets and the emergence of financial bubbles and crashes. These are driven by traders' herd mentality and are sudden but not wholly unexpected. At the same time, while informed participants can observe the build-up and emergence of bubbles, prediction of crashes - the tipping points - is non-trivial due to complex nonlinear interactions and the change in behaviour of market participants if the existence of a bubble becomes known. In collaboration with an industrial partner providing risk services (risk modelling, consulting, and management) to institutional clients, we aim to understand the crucial mechanisms that govern the class of complex systems represented by financial markets and the likelihood of regime changes in their behaviour. In order to achieve our objective, we will use the models and tools developed by statistical physics such as the models created to describe seismic activity and build novel extensions of these to account for the typical feedback mechanisms present in financial markets; in doing so, we aim to develop new characterisations and measurements of financial regimes and transitions. This work will directly inform the next generation of early warning risk systems developed by our industrial partner: such systems are used, for example, by the pensions industry and asset advisors to protect capital. It will also result in models (or methods of their analysis) that are applicable to broader classes of natural complex systems and their critical transitions.