Cross Model Sentiment Analysis in Stock Market: A Hybrid Approach Using Classical and Deep Learning Models

Cross Model Sentiment Analysis in Stock Market: A Hybrid Approach Using Classical and Deep Learning Models

Ashmit Gupta

INTRODUCTION

Background

In the dynamic and often volatile world of financial markets, investor sentiment plays a pivotal role in driving short-term price fluctuations. The rise of social media platforms, online financial forums, and real-time news dissemination has transformed the traditional landscape of stock market analysis. These digital channels have become rich sources of unstructured, sentiment-laden text data that reflect the emotions, perceptions, and expectations of investors. Harnessing this data to forecast stock price movement has emerged as a key focus area in financial analytics.

Sentiment analysis, a subfield of Natural Language Processing (NLP), is increasingly employed to extract emotional cues from financial texts. With the integration of Machine Learning (ML) and Deep Learning (DL) models, this process has evolved into a sophisticated tool for financial forecasting. However, while many models have been tested individually, there remains limited empirical research comparing classical ML models such as Support Vector Machines (SVM) and Naïve Bayes with DL architectures like LSTM (Long Short-Term Memory) and BERT (Bidirectional Encoder Representations from Transformers) in a hybrid ensemble framework.

This study aims to fill this gap by conducting a comprehensive comparative analysis and proposing a hybrid sentiment analysis model that leverages the strengths of both ML and DL techniques for stock market prediction.