ESCAPE - Efficiency and Spreading of the Financial Crisis and Algorithmic Policy Evolution (05/2012-04/2015)

Scope of the project

ESCAPE is an acronym for Efficiency and Spreading of the financial Crisis and an associated Analysis of the financial Policy Evolution. The research concerns the exploration of financial text news through Machine Learning and Text Mining techniques as well as the analysis of the incidents concerning the evolving sovereign debt crisis in Europe. For this, we use Financial News Data from Thomson Reuters as a 'historical display' of the last years.
Our current research focuses the following aspects:

• Contagion Effects of the European Souvereign Debt Crisis (AC, TG).
• Topic Detection in Financial Text News (DK, CS).
• Sentiment Classification of Financial News (CS).

This research project is funded by the CORE Research Program of the Fonds National de la Recherche (FNR) Luxembourg for a duration of 3 years.

Team

ESCAPE is an interdisciplinary project between the MINE Research Group of the Department of Computer Science and Communication and the Research Group of Prof. Grammatikos, Luxembourg School of Finance.The current project members are:

• Andreas Chouliaras (Dept. of Finance)
• Theoharry Grammatikos (Dept. of Finance)
• Dimitrios Kampas (Dept. of Computer Science)
• Christoph Schommer (Dept. of Computer Science)

Temporary Members

• Roxana Bersan (2013-14)
• Raphael Sirres (2014)
• Ethem Ocakdan (2014)

Interdiscpiplinary Working Group

• We have established an Internal Working Group on Computational Finance. For more information, please contact Prof. Grammatikos and/or Prof. Schommer.

Publications

2015

• A. Chouliaras. High Frequency Newswire Textual Sentiment: Evidence from international stock markets during the European Financial Crisis (2015).

2014

• A. Chouliaras, T. Grammatikos. Extreme Returns in the European Financial Crisis (2014).
• D. Kampas, C. Schommer, U. Sorger. A Hidden Markov Model to detect relevance in financial documents based on on/off topics, ECDA 2014, Bremen

2013

• A. Chouliaras, T. Grammatikos. News Flow, Web Attention and Extreme Returns in the European Financial Crisis. Proceedings 26th Australasian Finance and Banking Conference (2013).
• D. Kampas, C. Schommer. Topic Classification in Thompson Reuters Financial Text News. Abstract. ECDA 2013, Luxembourg.
• M. Minev, C. Schommer. News Representation with Multi-Word Features. Abstract, European Conference on Data Analysis, Luxembourg, July 2013.
• R. Bersan, D. Kampas, C. Schommer: A Prospect on How to Find the Polarity of a Financial News by Keeping an Objective Standpoint. Position Paper. ICAART 2013, Barcelona, Spain.
• M. Minev. Quantification of Financial News for Economic Surveys. ICDMW, 2013.
• M. Minev, C. Schommer. Domain-driven news representation using conditional attribute-value pairs. In N. Ferro, editor, PROMISE Winter School 2013: Bridging between Information Retrieval and Databases, volume 8173 of Lecture Notes in Computer Science. Springer, 2013.
• M. Minev, C. Schommer, U. Schaefer, and T. Grammatikos. Feature Extraction and Representation for Economic Surveys. In "Bridging between Information Retrieval and Databases", Bressanone, Italy, February 2013. (Best Poster Award). SIGIR Forum. June 2013, Vol. 47/1.

2012

• M. Minev, C. Schommer, T. Grammatikos. News and stock markets: A survey on abnormal returns and prediction models. Technical Report, August 2012.C. Schommer and M. Minev. Data Mining in Finance. Digital Enlightenment Forum, Seminar Trusted ICT for Finance, Luxembourg, April 2012.

Theses and Academic documents

• Dr. Mihail Minev : Feature Detection and Classification in Financial News. PhD Thesis. (2014)
• C. Schommer. Sentiment Barometer in Financial News. Course Report Machine Learning (Master of Science), 105pp.
• R. Sirres: A CSV to XML converter of TR news data (Bachelor thesis, Summer 2013). Supervison: (DK, CS).
• E. Ocakdam: Transforming unstructured data into structured data using NLP tools and techniques (Bachelor thesis, Summer 2013). Supervison: (MM, CS).

Scientific Contacts and Visits

• Dr. Stephen Clark, Dept. of Computational Linguistics, University of Cambridge, UK.
• Prof. Dr. Philip Treleaven, Dept. of Computer Science, University College London, UK.
• Prof. Dr. Yannis Ioannidis, Dept. of Computer Science, University of Athens, Greece.

Teaching

• Machine Learning - Top 10 Algortithms. 3rd semester course (MICS)
• Machine Learning - Sentiment Classificiation. 3rd semester course (MICS). Winter 2013/14.

Miscellaneous

There are some possibilities to join the team, either through PhD/PostDoc scholarships, offered by the Fonds National de la Recherche - see AFR (there are two calls per year, one in September and one in March, respectively) and/or by Master/Bachelor Thesis Projects and student assistantships. Feel free to contact us!

Acknowledgement

This research project is supported by the Fonds National de la Recherche Luxembourg