Automatic Prediction of Stock Market Behavior Based on Time Series, Text Mining and Sentiment Analysis: A Systematic Review

Abstract

Predicting stock market behavior is a challenge that has been studied and presented several solutions in the literature. Due to technological advances, methodologies have emerged and allowed new approaches to this problem in recent years. Text mining and sentiment analysis have been widely applied in this area. On the other hand, classic solutions as time series analysis continue to be used alone or with new methods. There is still no literature review of the joint use of these methods. In this way, this study presents a systematic review with 57 selected papers using time series, text mining, and sentiment analysis applied to predict financial stock market behavior. Through this research, it was observed that the use of data from social media and internet sites is a compound source of information, providing a better prediction. However, the selection and combination of these data in a relevant way are still limitations found in the proposed models.