Sentiment analysis of financial news for predicting stock price trends using NLP techniques in fintech

Cherng, Jun Kai (2025) Sentiment analysis of financial news for predicting stock price trends using NLP techniques in fintech. Final Year Project, UTAR.

PDF Download (4Mb)

Abstract

The stock market is highly influenced by news and investor sentiment, making trend prediction both challenging and valuable. This project develops a framework for stock price trend prediction by integrating sentiment analysis of financial news with historical market data. News headlines are cleaned and analyzed using VADER, TextBlob, BERT, and FinBERT to generate sentiment scores, which are merged with OHLCV price data and enriched with lagged returns and time-based features. Five machine learning models — Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), XGBoost, and LightGBM — are trained and tuned using walk-forward cross-validation. Their performance is evaluated using accuracy, precision, recall, F1-score, and confusion matrices, with XGBoost achieving the best results. Finally, a Power BI dashboard is built to visualize sentiment trends, market data, and model predictions, making insights interactive and actionable. Results show that incorporating sentiment features improves predictive performance, supporting data-driven decision-making for investors and analysts.

Actions (login required)