Removing systematic patterns in returns in a financial market model by artificially intelligent traders

Professorship/Faculty: Wissenschaftliches Institut für Hochschulsoftware der Universität Bamberg (ihb) 
Authors: Witte, Björn-Christopher
metadata.dc.contributor.contributor: Stübben, Felix
Corporate Body: BERG (Bamberg Economic Research Group)
Publisher Information: Bamberg : opus
Year of publication: 2012
Pages / Size: 34 S. : Ill., graph. Darst.
ISBN: 978-3-931052-92-8
Series ; Volume: BERG working paper series  ; 82
Source/Other editions: zuerst erschienen im BERG-Verlag, 2011
QK 622
Year of first publication: 2011
Language(s): English
Licence: German Act on Copyright 
URN: urn:nbn:de:bvb:473-opus4-24227
Document Type: Workingpaper
The unpredictability of returns counts as a stylized fact of financial markets. To reproduce this fact, modelers
usually implement noise terms − a method with several downsides. Above all, systematic patterns are not
eliminated but merely blurred. The present article introduces a model in which systematic patterns are removed
endogenously. This is achieved in a reality-oriented way: Intelligent traders are able to identify patterns and
exploit them. To identify and predict patterns, a very simple artificial neural network is used. As neural network
mimic the cognitive processes of the human brain, this method might be regarded as a quite accurate way of how
traders identify patterns and forecast prices in reality. The simulation experiments show that the artificial traders
exploit patterns effectively and thereby remove them, which ultimately leads to the unpredictability of prices.
Further results relate to the influence of pattern exploiters on market efficiency.
SWD Keywords: Kapitalmarkt ; Künstliche Intelligenz ; Mehragentensystem ; Online-Publikation
Keywords: financial markets; autocorrelations; artificial intelligence; agent-based modeling
DDC Classification: 330 Economics 
Release Date: 18. December 2012

File SizeFormat  
BERG82WitteseA2.pdf791.7 kBAdobe PDFView/Open