LSTM and NLP Based Forecasting Model for Stock Market Analysis

Abstract

The stock market prices change everyday by market forces (supply and demand). In recent years stock market forecasting becomes a successful approach to predict stock prices. Investors are investing in the stock market based on certain predictions. For guiding stock market investors, this research paper proposes a NLP and LSTM based forecasting model for stock market. According to our research, we found a strong bond between social media news and historical data. An algorithm is proposed for sentimental analysis to establish the correlation between the stock market values and the sentiments in news feed. In our model we have utilized two different methods Natural language processing (NLP) for the feature extraction and Long Short-Term Memory (LSTM) for training our dataset. The simulated experiment based performance evaluation and comparison of the proposed model outperformed the state-of-the-arts by achieving high prediction accuracy by reducing the mean square error up to 0.062.