Sentiment and Crisis in Financial Texts

Two talks blending together textual analysis and finance with applications to sentiment and crisis prediction.

#1 The Financial Crisis and the Financial Literature - a Textual Analysis

• by Dr. Alon Raviv, The Graduate School of Business Administration, Bar-Ilan University*

#2 A New Method for Sentiment Analysis of Financial Documents

• by Danny Lesmy, School of Business Administration, Hebrew University *

DETAILS

#1 The Financial Crisis and the Financial Literature - a Textual Analysis

• by Dr. Alon Raviv, The Graduate School of Business Administration, Bar-Ilan University*

In this talk I plan to review the causes and the remedies for the 2007-2009 financial crisis according to the different strands of the economic and financial literature. I will show how the academic field of finance and economic react to the current financial crisis and how this reaction differ from past crises. I will present preliminary results which are based on textual analysis as well as future directions and previous works.

Bio:
Dr. Alon Raviv is a Senior Lecturer at the Department of Business Administration, Bar Ilan University. His main areas of interest include:
Management, Banking, Corporate Finance and Financial Stability.
https://mba.biu.ac.il/en/raviv

#2 A New Method for Sentiment Analysis of Financial Documents

• by Danny Lesmy, School of Business Administration, Hebrew University *

Sentiment analysis refers to the process of determining the polarity (positive, negative or neutral) of a given text. The classical approach for sentiment analysis consists of creating a dictionary of tagged words used to map the sentiment of the text. This method has many disadvantages and limitations. For example, it does not consider how words are combined in a sequence and the order of the words in a sentence, e.g., positive words can turn to have a negative sentiment when they are preceded by a "not". Only recently, sentiment analysis research has begun to use advanced algorithms from Machine and Deep Learning algorithms to find the dependency / constituency parsing of the sentence. These algorithms are time consuming and require advanced computational resources, making them difficult to implement.
Specifically, in the financial literature, due to the complex structure of financial reports, the large number of sub-items and the technical domain-specific jargon, the use of advanced NLP algorithms is limited and sentiment analysis is mainly performed using the classical approach. In my talk I will present a new hybrid approach we have developed for sentiment analysis of financial documents, combining classic methods and advanced tools to achieve high performance. I will also discuss the multiple advantages of this new technique, and present interesting real-world results and financial insights.

Bio:

Danny Lesmy is a Ph.D. student at the Jerusalem School of Business and Administration at the Hebrew University of Jerusalem, under the supervision of Prof. Lev Muchnik. Danny’s Ph.D. deal with developing a new metric to measure text complexity and applying it to financial reports.