Investor Sentiment based Predictive Stock Market Forecasting using Deep Learning

To improve stock price prediction, the MS-SSA-LSTM model introduced in the Sentiment Analysis article of East Money forum post was integrated into this model. This model combines data from multiple sources, sentiment analysis, swarm intelligence algorithms, and deep learning techniques. It creates a special sentiment dictionary and provides a sentiment index, which allows insightful analysis of how market sentiment affects stock prices. The LSTM hyperparameters are fine-tuned using "Sparrow Search Algorithm (SSA)" to maximize the prediction accuracy. Experimental results highlight that the MS-SSA-LSTM model performs well. It is the perfect tool for accurate stock price prediction. The model is designed for turbulent financial markets and excels at short-term stock price prediction, providing ideas for dynamic decision-making by investors. Moreover, a hybrid LSTM + GRU model was provided to classify stock price sentiment. Moreover, a powerful ensemble approach was used with voting classifiers "(AdaBoost + Random Forest) for sentiment analysis and voting regressors (Linear Regression + Random Forest Regressor + K Nearest Neighbor Regressor)" for stock price prediction. These instruments fit current models (MLP, CNN, LSTM, MS-LSTM, MS-SSA-LSTM), thus improving the overall prediction performance. The development of an easy-to-use Flask framework with SQLite support simplified the registration, login, and model evaluation procedures, making it easier for users onboarding and testing.