Building a Sentiment Indicator Index for Investment Assets based on Natural Language Processing

Abstract:

The market for investment assets is subject to constant fluctuations due to forces interacting in a dynamic market environment. Market participants constantly analyze the current situation to optimize their purchase decisions. Proper prediction of market behavior based on market data maximizes investor's profit or minimizes loss. A key aspect of trading is to make assumptions of the future asset price. The forecast could benefit from a variety of techniques. One of them is to analyze an actual situation of an asset (Fundamental Analysis). Another is considering multiple statistical methods to predict price movements (Technical Analysis). Other methods interpret various pieces of publicly available information. This article focuses on the automatic sentiment analysis of statements made by market participants within a forum of individual investors. The research aim is to evaluate the sentiment towards a given asset. The research subjects are companies listed on the Warsaw Stock Exchange which are components of the indices WlG20, mWlG40, and sWlG8O.