LSTMSA: a novel approach for stock market prediction using LSTM and sentiment analysis

Abstract

Stock market prediction is one of the most popular use cases for machine learning models. A general model that can predict the rise and fall of stocks is an arduous task as there maybe multifarious factors that can affect stock prices. This paper attempts to create a model by emulating the approach traders, investors and analysts take to evaluate stock investment strategy. A conjunction of both technical analyses using available numerical data about stocks and fundamental analysis using news headlines are attempted to understand and predict market behavior for the Google stock. For this purpose, sentiment analysis is used to understand news data regarding the stock along with existing time series data as input for an LSTM neural network. It is observed that such an approach yields a more intuitive and accurate yet generalized model that can be used for prediction of the stock market.