Developing stock sentiment rank metric for forecasting stock sentiments: a statistical perspective

Abstract There has been a recent increase in the focus on sentiment research in the financial and stock market sectors. Large quantities of unstructured data that can be analyzed with the assistance of this technology include news stories, social media posts, financial records, and general opinions about individual organization. Researchers suggest a wide range of review articles to carry out these tasks, as every model differs in its internal and external operating characteristics. A comprehensive evaluation of sentiment analysis methodologies, including hybrid models, machine learning models, and lexicon-based models, is offered to facilitate the model selection process. From 2019 to 2024, 57 research articles were selected for review. The research offers an overview of models based on scalability metrics, assessment delay, deployment costs, implementation of complexity, and accuracy. This paper delves further into the limitations and complexities of sentiment analysis in the stock market, offering suggestions for future research to tackle these issues. A unique Stock Sentiment Rank Metric (SSRM) is developed for getting the insights regarding market psychology, behaviour of the investors and price movements. SSRM investigate the sentiment analysis methodologies more completely rather than just providing the sentiment score. SSRM provides a standardized, quantitative comparison of different models, ensuring objective evaluation. Qualitative comparisons alone can be subjective and inconsistent across different datasets and market conditions. This statistic will help readers find the best successful models for high-scalability situations by one metric for both performance and efficiency-oriented factors.