Relation-based Deep Learning for Finance

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering

Final Year Thesis Oral Defense

Title: "Relation-based Deep Learning for Finance"

by

NG Zhi Yong Ignavier

Abstract:

In the financially volatile market, stock trading is one of the most 
important exertions in finance and business to gain profit. To improve 
trading performance, the financial industry has been continuously applying 
sophisticated techniques in search of robust stock prediction model. 
However, stock predictions are typically performed on a discrete basis 
which only utilize the historical price of a given stock to predict its 
price in the future. In this thesis, we propose to leverage the relations 
between different stocks to improve the performance of stock prediction 
such that historical prices of highly correlated stocks are taken into 
account when predicting the future price of the stock itself. We conduct 
experiments to: (1) show that this could be achieved by using graph-based 
deep learning models (2) investigate which type of stock relation has 
higher predictive power.


Date            : 2 May 2019 (Thursday)

Time            : 16:20 - 17:00

Venue           : Room 5510 (near lifts 25/26), HKUST

Advisor         : Prof. KWOK Tin-Yau James

2nd Reader      : Dr. WONG Raymond Chi-Wing