The Impact of Machine Learning on Financial Forecasting and Market Trends

Abstract

The use of machine learning techniques in financial forecasting has improved the precision and speed of predicting market trends. This research explores the application of a feature-based approach using FinBERT, which leverages sentiment analysis of financial news to enhance forecasting reliability. The model was evaluated against benchmark machine learning models using a pipeline consisting of web scraping, data cleansing, sentiment labeling, and financial domain fine-tuning. Assessing the model and forecasting error rates revealed that the models built with FinBERT outperformed the others, thanks to their ability to evaluate and exploit finer sentiment metrics. This proved that models trained using specialized texts significantly decreased prediction errors, thus highlighting the importance of specialized domain language models. This study validates the capability of transformer-based structures in real-time market prediction to enhance the accuracy and reliability of AIpowered financial analysis, illustrating the continuing evolution of AI-powered financial analysis.

Access to Document

Other files and links

• Link to publication in Scopus

Fingerprint