A New Hybrid Model ARFIMA-LSTM Combined with News Sentiment Analysis Model for Stock Market Prediction

Abstract

We have used the advantages of both data-driven and news-driven models and came up with a novel hybrid model. This model combined the data-driven ARFIMA-LSTM with the NEWS SENTIMENT analysis model. So, in order to obtain a better and more accurate prediction at each time scale, we combined the capabilities of a data-driven model (ARFIMA-LSTM) with a news sentiments-driven model. This model outperforms the traditional models, such as SVM, Random Forest, ARIMA, KNN, GRNN, and LSTM, and it accurately forecasts the market.