Machine Learning Models for Portfolio Management Using Attention-based Sentiment Analysis

Abstract

Investor sentiment is a crucial factor that can impact the performance of a portfolio, as it can influence the buying and selling decisions of market participants. The growing trend of using machine learning (ML) in finance, particularly in portfolio management, to analyze large amounts of data quickly and accurately, identify patterns and trends, and generate insights that can inform investment decisions. One promising ML application in portfolio management is attention-based sentiment analysis, which uses attention mechanisms to identify the most relevant information for sentiment analysis. The thesis proposes attention-based sentiment analysis for portfolio management, which has the potential to revolutionize the investment industry by providing investors with more accurate and efficient investment decisions. The thesis aims to ansv, rer three research questions about the benefits, challenges, and implications of using attention-based sentiment analysis in portfolio management. The methodology involves gathering relevant financial news and social media data, developing and evaluating machine learning models incorporating attention-based sentiment analysis, and testing the models' performance. The study aims to shed I ight on the potential benefits or incorporating attention-based sentiment analysis into portfolio management models and the potential challenges and limitations of this approach. Ultimately, the insights gained from this study could help investors and portfolio managers develop more accurate and efficient investment strategies, leading to better investment outcomes.

Date of Award	20 Jul 2024
Original language	American English
Supervisor	IBRAHIM Elfadel (Supervisor)