Компаративний аналіз підходів на основі GPT та класичних методів вибору інвестиційного портфеля

Abstract

The thesis explores and compares two approaches to investment portfolio con struction: classical models (such as Markowitz optimization and Fama-French multi factor models) and a novel method based on large language models (LLMs), specifi cally GPT-3.5. The work applies machine learning techniques, natural language pro cessing (NLP), and mathematical modeling for data-driven financial decision making. The proposed solution includes the design and implementation of a GPT powered chatbot for portfolio advisory, with integration of FastAPI, AWS, and React technologies. The work evaluates the effectiveness of GPT-based recommendations versus traditional quantitative models using historical stock market data. Results in dicate that LLM-based systems can provide competitive and interpretable investment suggestions.