Title: Does web anticipate stocks? Analysis for a subset of systemically important banks
Authors: NARDO MichelaVAN DER GOOT Erik
Publisher: Publications Office of the European Union
Publication Year: 2015
JRC N°: JRC93383
ISBN: 978-92-79-44708-2
ISSN: 1831-9424
Other Identifiers: EUR 27023
OP LB-NA-27023-EN-N
URI: http://publications.jrc.ec.europa.eu/repository/handle/JRC93383
DOI: 10.2788/14130
Type: EUR - Scientific and Technical Research Reports
Abstract: Is web buzz able to lead stock behavior for a set of systemically important banks? Are stock movements sensitive to the geo-tagging of the web buzz? Between Dec. 2013 and April 2014, we scrape about 4000 world media websites retrieving all public information related to 10 systemically important banks. We process web news with a sentiment analysis algorithm in order to detect article mood. We show that web buzz does not seem to lead stock behavior as Granger test fails to support an average association that goes one-way from web to stocks. We nevertheless find a statistically sound anticipation capacity for single banks with gains ranging from 4 to 12%. Hierarchical clustering and Principal Component Analysis suggest that Euro area level decisions/facts do in fact drive stock behaviour, while web news about single banks only episodically make a difference in stock movements. Our analysis confirms that the location of the web source matters. The use of sources with international echo eliminates some of the noise introduced by irrelevant texts at the country level and improves the predictive power of the model up to 27.5%.
JRC Directorate:Space, Security and Migration

Files in This Item:
File Description SizeFormat 
jrc-report_fedra_final.pdf3.44 MBAdobe PDFView/Open

Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.