Stock Market Prediction Using Hard and Soft Data Fusion

The stock market fluctuates a lot due to economic factors and public sentiments. This challenge is exacerbated by the high volatility of stock price trends. To predict the trend of this market, better and more accurate forecasting is expected by combining different sources. We use the available historical data in three values using a Stacked LSTMs network. Then we achieve better results in network output by using OWA methods. The next step is to get the correct news from the website and use the Natural Language Toolkit (NLTK) to analyze their negative or positive impact on the market. Then, by using the technical knowledge and experience of several human experts, including the recognition of patterns in technical analysis and knowledge of the market news, we obtain the soft outputs, which finally combine the outputs of human experts with the outputs obtained from the network. One of the applications of combining hard and soft data is stock market forecasting, and finally, the proposed model for this work is presented.