Dynamic instabilities from information annihilation in speculative markets

• Applicant: Dr. Felix Patzelt
• Subject Area: Statistical Physics, Nonlinear Dynamics, Complex Systems, Soft and Fluid Matter, Biological Physics
• Term: from 2016 to 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 283620475

Financial markets exhibit extreme price fluctuations. Their statistics follow scaling-laws which are nearly universal across different markets and time periods. The underlying causes for these "stylised facts" are unknown. (Neo-)classical economic theories only allow for significant fluctuations due to external information entering the market. That is, "news". Otherwise, markets should be in equilibrium. Many empirical studies, in contrast, concluded that most large price changes cannot be attributed to external news.The "stylised facts" are often associated with the "critical phenomena" known from physics. So far, however, they cannot be fully explained from first principles shared across different markets. Instead, a zoo of out-of-equilibrium models with different mechanisms and shortcomings has been developed over recent years.A possible first step towards a more fundamental understanding of the "stylised facts" was taken with an abstract model demonstrating that speculative markets which absorb predominantly self-generated information may exhibit both: evolution towards information-efficient equilibrium states as well as their subsequent destabilisation. These dynamics reproduce the empirical scaling of log returns, which quantify price changes, and their long-ranged temporal clustering.While this abstract theory has the potential to reconcile conflicting views on the state of operation of financial markets - information efficiency and complex out-of-equilibrium dynamics - its real-world relevance is unknown. Hence I aim to transfer the theory onto a tangible description of real high-frequency trading. I will collaborate with Jean-Philippe Bouchaud, a renown researcher in statistical physics and finance. He is also chairman of Capital Fund Management (CFM), which will provide high-quality data that will serve to develop and test realistic models.I hypothesise that the "stylised facts" follow from general properties of a dynamical balance of the impacts of opposing market forces. I will therefore investigate also information transfer over different time-scales and possible generalisations of the theory. I aim to develop a consistent framework for describing financial markets across different levels of abstractions in multi-agent models, and up to macroscopic descriptions in the form of diffusion processes.Moreover, I hypothesise that a general principle can account for similar phenomena in many complex adaptive systems that feature an adaptive balance of opposing forces. One instantiation of this principle is human stick-balancing where absorption of predictable trends can reduce mean fluctuations, but also increase the risk for rare, extreme errors. Together with Klaus Pawelzik, I previously demonstrated this principle in a motor control model and in human balancing experiments. Therefore, we plan to collaborate on the identification of system-independent features of this Information Annihilation Instability (IAI).

• DFG Programme: Research Fellowships
• International Connection: France
• Host: Professor Dr. Jean-Philippe Bouchaud