High-Frequency Trading based on Limit Order Book via Machine Learning

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

Final Year Thesis Oral Defense

Title: "High-Frequency Trading based on Limit Order Book via Machine 
Learning"

by

MANGLA Aditya Vikram

Abstract:

High-frequency trading (HFT) has revolutionized financial markets, with 
algorithms now executing the majority of trades across global exchanges. This 
final year thesis explores the application of machine learning techniques to 
HFT strategies using limit order book (LOB) data, with a particular focus on 
market making rather than market taking to capitalize on rebate 
opportunities. The research develops a highly optimized backtesting engine 
capable of efficiently processing vast quantities of LOB data, and implements 
a series of increasingly sophisticated market-making strategies. Beginning 
with naive baseline approaches, the project iteratively develops more complex 
models including grid-based trading, Avellaneda-Stoikov skew-adjusted 
methods, and the Gueant-Lehalle-Fernandez-Tapia (GLFT) framework. The 
culmination of this progression is the application of reinforcement learning 
techniques to dynamically optimize strategy parameters. Performance 
evaluation across cryptocurrency markets demonstrates the challenges and 
potential of machine learning in HFT. This research establishes a foundation 
for future work in this rapidly evolving field and highlights the technical 
complexities involved in developing profitable HFT strategies.


Date            : 29 April 2025 (Tuesday)

Time            : 14:00 - 14:40

Venue           : Room 5504 (near lifts 25/26), HKUST

Advisor         : Prof. SANDER Pedro

2nd Reader      : Dr. CHEN Qifeng