Analysis of Effect of Historical Prices And News on The Stock Market

Abstract

The proliferation of data usage and insights gained from it has seen its applications manifold in the recent decades. One such application of data available is through analysis of the effect of news and historical data on the stock market. The paper has analysed the effect of historical prices and news on the Stock market using various models. Since news and historical data can gravely impact the Stock market, the paper uses Natural Language Processing (NLP) to analyse the news and several predictive analysis models to make use of the historical data to make predictions concerning the stock prices. The NLP algorithm utilized makes use of the text as input and tokenizes each sentence into words to make the process of analysis easier. Google's Bidirectional Encoder Representations from Transformers (BERT) model is used for the analysis of the news. For the historical data analysis, the two models namely Auto-regressive Integrated Moving Average (ARIMA) and Long short-term memory (LSTM) have been compared and post evaluation it has been found that LSTM Recurrent Neural Networks (RNN) is a superior model. The data for these models, has been procured from the Internet to ensure relevancy and credibility. Both the models used in the analysis of news and historical data have been cross-validated several times which has yielded adequate accuracy.