Stock Ranking Analysis using AI

Abstract

Financial markets are embedded and have had significant impact on many areas such as business, jobs, education and technology, which all have an impact on the economy. It is an extremely challenging field to accurately predict because stock market movements and price behavior are hard to analyze given its dynamic, nonstationary, nonparametric, nonlinear, and chaotic nature and they are impacted by several interrelated factors such as economic, political, psychological, and company-specific variables [1]. There is a rich history of stock prediction using analytical and traditional statistical tools such as Moving Average, Auto-regressive Integrated Moving Average, and Vector Auto-regressive Moving Average. But recently, we have the emergence of advanced Machine Learning methods that provide an entirely new toolbox that could be used to address complex prediction problems. This research will focus on developing Predictive Equity Ranking (PER) methodologies using modern machine learning tools and Learning to Rank (LTR) algorithms. The dataset that would be used is a real-world dataset (S&P500). The expected outcome is a stock ranking based on the most profitable stocks. By succeeding in doing so, a portfolio can be further constructed with the highest revenue generating stocks, for both long and short strategies.

Date of Award	Apr 2023
Original language	American English
Supervisor	PANAGIOTIS Liatsis (Supervisor)