Investors and financial analysts normally rely mostly on quantitative data and reading annual reports with qualitative information when analyzing the company's success. This task aims to examine the usefulness of computer text processing, accelerate the process and facilitate a more profound analysis of multiple companies.
To answer this question, we analyse the annual reports of the companies listed on the Ljubljana Stock Exchange. Several text features are computed, like subjectivity, sentiment, display of eight emotions, certainty and readability. The study compares various features of English and Slovene texts and also evaluates whole reports versus their crucial sections, which comprise strategy, vision, and mission. Finally, the connection between text attributes and financial indicators is explored.
One of the findings is that the absolute values of correlations are usually higher on English than on Slovene reports, as well as on the whole reports than only on their crucial sections. Indicators with the most correlations are debt to liabilities ratio and days of supply. Correlations with return on equity and return on assets in the next year almost never exceed the weak correlation.
|
