Understanding the Market Trends: A Hybrid Approach to Stock Price Prediction Using RNNs and Transformer-Based Sentiment Analysis
Understanding the Market Trends: A Hybrid Approach to Stock Price Prediction Using RNNs and Transformer-Based Sentiment Analysis

Summary/Abstract: Stock price prediction is a critical yet challenging task in financial markets due to the complexity and volatility of asset movements. This paper presents a hybrid approach that combines Recurrent Neural Networks (RNN), particularly Long Short-Term Memory (LSTM) models, for time-series prediction with Transformer-based text analysis to capture sentiment from financial news. The study focuses on predicting Apple Inc.'s (AAPL) stock price, using three years of historical data alongside news sentiment analysis. The LSTM model captures temporal dependencies in the stock prices, while the Transformer model extracts relevant features from unstructured textual data, offering insights into market sentiment and external events. The results demonstrate that integrating sentiment data with stock price predictions significantly improves model accuracy, as reflected by a reduction in mean squared error (MSE) compared to models based solely on price data. This hybrid model offers a more holistic approach to financial forecasting, combining quantitative and qualitative data for enhanced prediction. The paper contributes to the field of machine learning in finance by highlighting the benefits of hybrid modelling approaches, and it opens avenues for future research on broader applications in other asset classes and more diverse data sources.

• Details
• Contents