Downloads: 2
Research Paper | Finance | Volume 15 Issue 6, June 2026 | Pages: 1073 - 1081 | India
Comparative Analysis of NLP Frameworks FinBERT & VADER in Predicting Market Sentiment of NIFTY 50
Abstract: Financial news headlines constitute a critical source of real-time information for investors, analysts, and policymakers operating in equity markets. Sentiment analysis of such headlines provides valuable insights into market psychology and supports data-driven investment decision-making. This paper presents a comparative analysis of two sentiment analysis frameworks, namely Valence Aware Dictionary for sEntiment Reasoning (VADER), a lexicon-based model, and FinBERT, a domain-specific transformer model pre-trained on financial corpora, applied to financial news headlines pertaining to the NIFTY 50 index of the National Stock Exchange (NSE) of India. Headlines were collected from leading Indian financial news platforms and manually classified into three sentiment categories: positive, negative, and neutral, from a financial perspective. Following standard text preprocessing procedures including tokenization, stop word removal, and lemmatization, the headlines were processed through both models. Model performance was evaluated using four metrics: accuracy, sensitivity, specificity, and neutral specificity. Furthermore, this study investigates the challenges encountered by both models when processing Indian financial terminology, market-specific abbreviations, and the linguistic nuances characteristic of NIFTY 50 news headlines. Statistical hypothesis testing is employed to validate the findings. Experimental results reveal that neither model demonstrates uniform superiority across all evaluation metrics. FinBERT achieves higher accuracy and sensitivity, reflecting its capacity to leverage contextual understanding derived from domain-specific pre-training, while VADER demonstrates comparatively stronger performance in neutral specificity, attributed to its structured lexicon-based classification approach. These findings underscore the importance of metric-specific model selection in financial sentiment analysis applications.
Keywords: Sentiment analysis, VADER, FinBERT, NIFTY 50, Financial news
How to Cite?: Anjani Shivraj, "Comparative Analysis of NLP Frameworks FinBERT & VADER in Predicting Market Sentiment of NIFTY 50", Volume 15 Issue 6, June 2026, International Journal of Science and Research (IJSR), Pages: 1073-1081, https://www.ijsr.net/getabstract.php?paperid=SR26619135658, DOI: https://dx.doi.org/10.21275/SR26619135658