Wykorzystanie analizy tekstu do oceny zachowania stóp zwrotu indeksu WIG20
Using text analysis for evaluating the behaviour of ra- tes of return from the WIG20 index

Summary/Abstract: The aim of the article is to indicate the possibility of using text analysis for the research of dynamics of the Polish capital market. The first part of the article notes the changes which took place within the data market in recent years and their impact on the discounting of information by stock market investors. The Information Effectiveness Hypothesis and the paradigm of behavioural finance are the basis of theoretical considerations. The second part presents the result of a study, the objective of which was to build an algorithm allowing the prediction of the WIG20 index rates of return based on text data. The test sample consisted of 901 papers published in the “Parkiet” magazine and 748 daily rates of return. The study was conducted using algorithms processing natural language and decision trees used for classification. The results of the study allowed an indication of a model in which the precision and accuracy indicators exceeded a score of 50%.

• Details
• Contents