Machine learning in quantitative finance

Ha, Youngmin (2017) Machine learning in quantitative finance. PhD thesis, University of Glasgow.

Full text available as:
[thumbnail of 2017haphd.pdf] PDF
Download (3MB)
Printed Thesis Information: https://eleanor.lib.gla.ac.uk/record=b3289009

Abstract

This thesis consists of the three chapters.

Chapter 1 aims to decrease the time complexity of multi-output relevance vector regression from O(VM^3) to O(V^3+M^3), where V is the number of output dimensions, M is the number of basis functions, and V<M. The experimental results demonstrate that the proposed method is more competitive than the existing method, with regard to computation time. MATLAB codes are available at http://www.mathworks.com/matlabcentral/fileexchange/49131.

The performance of online (sequential) portfolio selection (OPS), which rebalances a portfolio in every period (e.g. daily or weekly) in order to maximise the portfolio's expected terminal wealth in the long run, has been overestimated by the ideal assumption of unlimited market liquidity (i.e. no market impact costs). Therefore, a new transaction cost factor model that considers market impact costs, estimated from limit order book data, as well as proportional transaction costs (e.g. brokerage commissions or transaction taxes in a fixed percentage) is proposed in Chapter 2 for both measuring OPS performance in a more practical way and developing a new OPS method. Backtesting results from the historical limit order book data of NASDAQ-traded stocks show both the performance deterioration of OPS by the market impact costs and the superiority of the proposed OPS method in the environment of limited market liquidity. MATLAB codes are available at http://www.mathworks.com/matlabcentral/fileexchange/56496.

Chapter 3 proposes an optimal intraday trading strategy to absorb the shock to the stock market when an online portfolio selection algorithm rebalances a portfolio. It considers real-time data of limit order books and splits a very large market order into a number of consecutive market orders to minimise overall transaction costs, consisting of market impact costs as well as proportional transaction costs. To be specific, it optimises both the number of intraday tradings and an intraday trading path for a multi-asset portfolio. Backtesting results from the historical limit order book data of NASDAQ-traded stocks show the superiority of the proposed trading algorithm in the environment of limited market liquidity. MATLAB codes are available at http://www.mathworks.com/matlabcentral/fileexchange/62503.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Keywords: relevance vector machine, online portfolio selection, algorithmic trading.
Subjects: H Social Sciences > HG Finance
Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Colleges/Schools: College of Social Sciences > Adam Smith Business School > Economics
Supervisor's Name: Cerrato, Prof. Mario and Sogiakas, Dr. Vasilios
Date of Award: 2017
Depositing User: Youngmin Ha
Unique ID: glathesis:2017-8558
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 03 Nov 2017 14:49
Last Modified: 17 Nov 2017 13:11
URI: https://theses.gla.ac.uk/id/eprint/8558
Related URLs:

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year