Title: Multi-agent financial market simulation: evolutionist approach
Authors: Badiâa Hedjazi; Mohamed Ahmed-Nacer; Samir Aknine; Karima Benatchba
Addresses: Information Systems Division, CERIST Research Centre, 5 Rue des Frères Aissou, Ben Aknoun, Algiers, Algeria ' Information Systems Laboratory, USTHB University, BP 32 EL ALIA 16111, Bab Ezzouar, Algiers, Algeria ' GAMA Laboratory, Université Claude Bernard Lyon 1, Campus de la Doua, Batiment Nautibus, 43, bd du 11 novembre 1918, 69622 Villeurbanne cedex, France ' ESI, National High School of Computer Science, BP 68M Oued-Smar, 16309, El-Harrach, Algiers, Algeria
Abstract: Financial markets are complex systems consisting of entities interacting and evolving in an uncertain environment. Their modelling and simulation requires the use on the one hand of a suited technology that is multi-agent systems (MASs) to model the various actors of a market, and on the other hand the evolutionary game theory to formalise interactions and heterogeneous investment strategies. The goal of this paper is to model, simulate and analyse financial markets dynamics. For this purpose, we propose three market models (fundamentalist, strategic, conventionalist) summarising various facets of real market speculation depending on the information held and the price formation process chosen by the investors. Each model is built using a multi-agent system. Moreover, investors' agents are modelled by classifier systems that are advanced structures to study their evolutionary and adaptive aspects.
Keywords: financial markets; multi-agent systems; MAS; simulation; evolutionary game theory; classifier systems; agent-based systems; modelling; investment strategies; market dynamics; market speculation.
DOI: 10.1504/IJSPM.2013.057538
International Journal of Simulation and Process Modelling, 2013 Vol.8 No.2/3, pp.185 - 199
Received: 11 May 2011
Accepted: 10 May 2012
Published online: 29 Jul 2014 *