Artificial Intelligence in Systematic Trading

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

Final Year Thesis Oral Defense

Title: "Artificial Intelligence in Systematic Trading"

by

LIU Xin

Abstract:

Systematic trading enjoys multiple advantages over the more traditional 
discretionary trading. To improve trading performance, the industry has 
been relentlessly applying new advanced techniques in search of 
successful systematic strategies. A recent trend of systematic trading 
research is to incorporate artificial intelligence, or more 
specifically, machine learning. Among the three major machine learning 
paradigms, reinforcement learning is found to be the most suitable model 
for the systematic trading problem. This research attempts to improve on 
previous reinforcement learning applications in systematic trading by 
incorporating a RNN layer with LSTM cells into the structure. The new 
reinforcement learning structure is termed Memory-based Recurrent 
Reinforcement Learning (MRRL). The performance of MRRL is compared with 
the most canonical reinforcement learning trading structure, Direct 
Recurrent Reinforcement Learning (DRRL) and it is shown that MRRL can 
achieve a higher profit and a more robust Sharpe ratio across multiple 
major U.S. stocks on a minute-level dataset.

Date            : 25 April 2018 (Wednesday)

Time            : 18:00 - 19:00

Venue           : Room 1505 (near lifts 25/26), HKUST

Advisor         : Prof. LEE Dik-Lun

2nd Reader      : Dr. LEUNG Wai Ting