USING NLP TO QUANTIFY THE EFFECTS OF NON-GAAP MEASURES TO PREDICT THE OUTCOME OF SECURITIES LAWSUITS
Date
2019-12-10T18:44:18Z
Authors
Taylor, Stacey Dianne
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The Management Discussion and Analysis (MD&A) is arguably the most tonal section of the reports provided to the U.S. Securities and Exchange Commission. As part of that dialogue, companies use non-standardized financial metrics known as non-GAAP measures that do not conform with Generally Accepted Accounting Principles. Our research presents a novel extractive approach using Sentiment Analysis to measure the impact that non-GAAP measures have on the common investor versus those who are financially savvy. We find that sentiment declines once the non-GAAP sentences have been extracted with a statistical significance at the p=0.01 level. Building on this, our second research question investigated if we could use a similar approach with machine learning to predict the outcome of securities class action lawsuits. We find that we are able to predict the aggregate outcome of the lawsuits with a recall of 0.9142 using the Random Forest classifier.
Description
Keywords
Natural Language Processing, Machine Learning, Finance, Accounting, Statistics, Sentiment Analysis