Stock Direction Prediction Using Sentiment Analysis of News Articles

Abstract

It is widely acknowledged that stock price prediction is a job full of challenges due to the highly unpredictable existence of financial markets. Many market participants or analysts, however, attempt to predict stock prices using different mathematical, econometric, or even neural network models in order to make money or understand the nature of the equity market. In the past few years, a lot of models based on deep learning have been gaining popularity for predicting the volatility of the stock market prices. In this paper, the outcomes of many classical deep learning models such as LSTMs, GRUs, CNNs, and their several common variants are contrasted with two distinct stock price prediction targets: absolute stock price and volatility. The aim of the comparative study is to find out which model is the best fit for stock market prediction. We also attempt to research the relationship between news and stock trends, believing that news stories have an impact on the stock market by incorporating sentiment analysis into our model. Our methodology was to scrape news articles of a particular stock and use the corpus gathered to generate a sentiment score which is further used as an input to the model.