Integrating EEMD and ensemble CNN with X (Twitter) sentiment for enhanced stock price predictions

This research proposes a novel method for enhancing the accuracy of stock price prediction by combining ensemble empirical mode decomposition (EEMD), ensemble convolutional neural network (CNN), and X (Twitter) sentiment scores based on historical stock data. The complexity and volatility of financial markets pose challenges to accurate stock price forecasting. To address this challenge, the presented approach utilizes EEMD to decompose the original stock price time series, X sentiment analysis data, and relative strength index (RSI) technical indicator data obtained from daily stock fluctuations into intrinsic mode functions (IMFs) and a residual component. Subsequently, an ensemble CNN is constructed, comprising parallel subnetworks that learn distinct IMF representations, and their combined predictions result in robust stock price forecasts. This ensemble CNN consists of multiple parallel subnetworks, each learning distinct IMF representations, and combining their predictions yields a robust stock price forecast. X sentiment scores are incorporated through a separate CNN that analyzes sentiment in tweets related to target equities, capturing polarity and intensity. Experiments with actual stock price and X data show that the proposed "EEMD–ensemble CNN" model outperforms baseline methods in accurate stock price forecasting. The incorporation of X sentiment scores improves forecasts by accounting for the influence of public sentiment on stock price fluctuations. This study demonstrates the potential benefits of social media sentiment analysis for financial forecasting and offers practical implications for investors, traders, and financial analysts seeking informed decisions in dynamic stock market environments.