The Battle of the Models - Modern Takes on Traditional and Machine Learning Techniques in Empirical Finance

Aug 3, 2026

Upload by Levi

A dissertation advances empirical finance research by contrasting traditional models with modern and machine learning approaches. The work focuses on measuring asynchronicity in financial time series using dynamic time warping, evaluating the persistence of the S&P index effect, and analyzing biases and overfitting in machine learning return predictions. It integrates validation through simulation, improved beta estimation, and evidence on global price discovery, then concludes with implications for model selection and robust inference in asset pricing.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages171

Reading Time7 h 11 min

AccessFree to download

Introduction

Asynchronicity between financial time series

Is the S&P index effect dead?

Biases and overfitting in cross-sectional machine learning models

Thesis outline

To lead or to lag? Measuring asynchronicity in financial time series using dynamic time warping

Validating dynamic time warping

A better beta

Global markets price discovery

Conclusion

The index effect is not dead, it has mutated

Introduction

Index changes sample construction

The Battle of the Models: Modern Takes on Traditional and Machine Learning Techniques in Empirical Finance Clint Howard A dissertation submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy Finance Discipline Group, UTS Business School University of Technology Sydney Principal supervisor: Associate Professor Vitali Alexeev Co-supervisor: Professor T¯alis J. Putni¸nˇs October 2023 Certificate of Original Authorship I, Clint Howard, declare that this thesis, is submitted in fulfilment of the requirements for the award of Doctor of Philosophy, in the Finance Discipline Group at UTS Business School at the University of Technology Sydney. This thesis is wholly my own work unless otherwise referenced or acknowledged. In addition, I certify that all information sources and literature used are indicated in the thesis. This document has not been submitted for qualifications at any other academic institution. This research is supported by the Australian Government Research Training Program. Production Note: Signature: Signature removed prior to publication. Date: October 11, 2023 Acknowledgments I would like to thank my supervisors, Vitali and Talis. I am extremely grateful for the guidance and research mentorship they both provided me throughout my Ph.D. Considering my slightly unorthodox pathway into the Ph.D. program at UTS, Vitali and Talis showed great patience with me as I balanced my full-time work commitments with the completion of my thesis. I am also thankful to the support and guidance of Christina Sklibosios Nikitopoulos, who was always available for technical support and ensuring that I met all the requirements for the Ph.D. program as well as accommodating my international move along the way. During my Ph.D. journey, I had the privilege of working for two asset management teams, Macquarie Systematic Investments and Robeco Quantitative Investments. I would like to thank my research colleagues, Samir Vanza and Steve Wong, from Macquarie. Their mentorship and research guidance had significant influence on my development as a researcher. I would also like to thank Ben Leung for his encouragement and supporting me to embark on a part-time Ph.D. program whilst continuing to work full-time. The completion of my Ph.D. has been significantly supported by many research colleagues at Robeco. I would like to specifically thank David Blitz and Harald Lohre, who provided comprehensive feedback, critiques, suggestions, and support for my research. I thank my parents for their everlasting encouragement of my life pursuits. Finally, I thank Pearl, whose support, patience, and understanding have kept me going throughout. Preface Chapters 2–4 of this thesis have been simultaneously developed as working papers. These working papers have been presented at academic conferences and to my employers during my doctoral candidacy (Macquarie Systematic Investments and Robeco Quantitative Investments) . The below list presents each working paper and the relevant presentations. 1. Howard, C., Putnins, T. and Alexeev, V. (2020), To lead or to lag? Measuring asynchronicity in financial time series using dynamic time warping, Working paper, UTS Business School. ❼ 2022 Robeco Institutional Asset Management Research Seminar Series. Rotterdam, The Netherlands. ❼ 2023 Lancaster University Management School Financial Econometrics Conference. Lancaster, United Kingdom. 2. Howard, C., Putnins, T. and Alexeev, V. (2021), The index effect is not dead, it has mutated, Working paper, UTS Business School. ❼ 2022 Robeco Institutional Asset Management Research Seminar Series. Rotterdam, The Netherlands. 3. Howard, C. (2023), Less is more? Biases and overfitting in machine learning return predictions, Working paper, UTS Business School. ❼ 2023 Robeco Institutional Asset Management Research Seminar Series. Rotterdam, The Netherlands. ❼ 2023 Inquire Europe Autumn Seminar. Colog

FAQs

What is the dissertation’s main research focus in empirical finance?

It examines how modern and machine learning techniques compare with traditional approaches when modeling financial markets and predicting returns.

How does the thesis measure asynchronicity between financial time series?

It uses dynamic time warping, supported by validation through simulation and subsequent empirical analysis.

What concerns about machine learning models does the dissertation address?

It studies biases and overfitting in cross-sectional machine learning models for return prediction and discusses how these issues affect reliability.

Aug 3, 2026

Upload by Levi