Multi-agent FX-market modeling based on cognitive systems

We present an approach of multi-agent market modeling on the basis of cognitive systems with three functionality features. These features are perception, internal processing and acting. A cognitive system is structurally represented by an error correction neural network. On the mirco-level we describe agents decisions behavior by combining cognitive systems with a framework of multi-agent market modeling. By aggregating agents decisions we axe able to capture the underlying market dynamics on the macro-level. As an application, we apply our approach to the DEM / USD FX-Market. Fitting real-world data, our approach is superior to more conventional forecasting techniques.