Dynamic Interaction in Financial Markets

Dynamic Interaction in Financial Markets

Following the May 6, 2010 Flash Crash there is has been increasing interest in the dynamic behaviour of financial markets. 

Given that the majority of exchange-based trading is computer-based, often with computers deciding how much to trade at what prices and at which venues, and given the rise of High Frequency Traders both as proprietary traders and as market makers, we are particularly interested in the dynamic behaviour of interacting financial algorithms.  This includes inter-alia market-making algorithms, benchmark algorithms, arbitrage algorithms, momentum algorithms, contrarian algorithms and the matching algorithms used in the exchanges.

Our research is proceeding along the following lines:

  1. To model dynamic interaction in discrete time.  We focus on discrete-time modelling because in reality all interaction between computers is  conducted in discrete time (albeit very fast discrete time, as data is latched into and out of the interfaces of digital computers), and because attempting to apply a continuous-time model to a discrete-time system can be problematic (cf Brigo and Mercurio 2000). 

  2. To model (i) detailed algorithms and strategies derived from published academic work; (ii) unpublished  algorithms derived from private communication and personal experience; and (iii) representative algorithms that distil certain characteristics of real algorithms.

  3. To explore the dynamics of simple dynamical systems by detailed specification of the associated discrete-time equations, by tracing the results of those interacting equations, and by analysing various state spaces of those systems.  Of particular interest are issues such as information delay and feedback networks.

  4. To explore the dynamics (including delay and feedback) of complex dynamical systems via simulation.  To this end we have developed a Simulator for Interaction Dynamics (our working name for this simulator is "InterDyne").  The simulator is currently used to simulate interaction dynamics in financial markets, but the InterDyne framework is potentially applicable to other domains as well.  The InterDyne Simulator supports precise control over communication topology and each communication link may have an associated information delay. 

Christopher D. Clack
Department of Computer Science
Gower Street