Portfolio Management with Reinforcement Learning

stock trading image

Introduction

The financial market has been democratized by ever advancing technology and easier access to financial data. The rise in machine learning has driven a trend of stock prediction and more technical analysis, especially facilitating short-term trading.

However, it is found that most of these analyses could not “beat the market” and people with limited knowledge in finance may overlook the hidden risk while pursuing high returns in their prediction and analysis. Small investors may only focus on a specific stock’s trend and assign all their capital into it. This could bring significant risk given the stock’s sensitivity to macroeconomic factors and isosymmetric risk induced by the firm of the stock itself.

Therefore, this project aims to apply artificial intelligence in portfolio management instead in a bid to take into consideration the downside risk of investment and produce a more consistent expected performance. It also introduces the importance of portfolio diversification as stated by Markowitz for a robust investment strategy.

Supervisor

Professor Huang, Zhiyi

Members

Chan Hin Ting
Tam Pai Lok
Mak Wan Sing