An intelligent text analytics approach to sentiment-based portfolio decision making in the electric vehicle market

Investment portfolio management increasingly requires the integration of financial fundamentals with market sentiment derived from rapidly expanding online information sources. The widespread availability of real-time financial news and opinion-based content has heightened the complexity of investment decision making, particularly in dynamic and innovation-driven sectors such as Electric Vehicles (EVs). Existing sentiment analysis research in finance typically rely on large-scale, general-purpose datasets that depend on extensive labeled corpora, with high-dimensional feature representations and are best suited for broad market contexts, and are difficult to replicate in specialized market segments where data availability is limited. The challenge of constructing an effective sentiment model from a small, domain-specific corpus remains largely unaddressed. This study proposes an intelligent text analytics approach for sentiment-based portfolio decision making that enhances and complements traditional financial analysis with systematically extracted textual information. Using natural language processing and computational linguistic techniques, we construct a compact yet informative market-specific lexicon set for sentiment measurement. Textual data are collected from Bloomberg and Yahoo Finance, comprising 2500 news articles published between January 2024 and August 2025 and covering two United States EV companies, Tesla and Rivian, and two Chinese EV companies, NIO and Li Auto. The proposed sentiment analytics framework, is evaluated using supervised machine learning classifiers to categorize market sentiment into positive, neutral, and negative signals that correspond to portfolio allocation decisions. Empirical results demonstrate strong classification performance, achieving F1-scores in the high eighty percent range across both training and testing datasets. The findings indicate that combining domain-specific textual features with data-driven analytics yields a nuanced understanding of market sentiment and emerging trends in the EV sector. Although focused on EVs, the framework is transferable to other market segments with minimal adaptation, offering financial analysts a systematic and scalable decision support tool for sentiment-informed portfolio management.