Forecasting Equity Prices using LSTM and BERT with Sentiment Analysis

Abstract

In the recent past, the utilization deep learning techniques to estimate stock prices has drawn a lot of research interest. This research study examines the application of BERT (Bi-directional Encoder Representations from Transformers) and LSTM (Long Short-Term Memory) for stock price forecasting utilizing historical stock data and sentiment classification of twitter posts. The proposed model combines the power of LSTM for capturing the sequential patterns in stock prices and the contextual understanding ability of BERT in analyzing tweet sentiment. The results of the empirical evaluation showed that in terms of accuracy and robustness, the proposed model surpassed conventional time-series models and other models based on deep learning. This study provides a proof-of-concept that the combination of LSTM and BERT can be a promising approach for stock price prediction. Furthermore, the study highlights the potential for future research to explore other deep learning techniques and data sources to improve the accuracy and robustness of stock price prediction models.