A Machine Learning Approach to Sector Based Market Efficiency

Aug 4, 2026

Upload by Anda

A machine learning study evaluates sector-based market efficiency using limit order book reconstructed data and sector representative comparisons. The work defines a research question, collects and preprocesses time-series financial data, and trains multiple neural models including dense, LSTM, and convolutional architectures. Model performance is analyzed with sector-level precision, hyper-parameter tuning, and precision adjustments used as a proxy for efficiency, followed by interpretation of results. The report concludes by summarizing findings and outlining discussion and further work directions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages31

Reading Time1 h 18 min

AccessFree to download

Acknowledgments

List of Tables

List of Figures

List of Acronyms

Chapter 1: Introduction and Related Work

Research Question

Background and Related Work

Chapter 2: Methodology

Data Sourcing and Collection

Data Prepossessing

Chapter 3: Results

Models

Hyper Parameters

Chapter 4: Conclusion

Chapter 5: Discussion and Further Work

References

Bowdoin College Bowdoin Digital Commons | Honors Projects | Student Scholarship and Creative Work | | --- | --- | | 2023
A Machine Learning Approach to Sector Based Market Efficiency
Angus Zuklie
Bowdoin College
Follow this and additional works at: [https://digitalcommons.bowdoin.edu/honorsprojects](https://digitalcommons.bowdoin.edu/honorsprojects)
Part of the Computer Sciences Commons | | Recommended Citation Zuklie, Angus, "A Machine Learning Approach to Sector Based Market Efficiency" (2023) . Honors Projects. 443. [https://digitalcommons.bowdoin.edu/honorsprojects/443](https://digitalcommons.bowdoin.edu/honorsprojects/443) This Open Access Thesis is brought to you for free and open access by the Student Scholarship and Creative Work at Bowdoin Digital Commons. It has been accepted for inclusion in Honors Projects by an authorized administrator of Bowdoin Digital Commons. For more information, please [contact mdoyle@bowdoin.edu](contact mdoyle@bowdoin.edu), [a.sauer@bowdoin.edu](a.sauer@bowdoin.edu). A Machine Learning Approach to Sector Based Market Efficiency An Honors Paper for the Department of Computer Science By Angus Zuklie Bowdoin College, 2023 © Angus Zuklie 2023 I believe the market accurately reflects not the truth, which is what the efficient market hypothesis says, but it accurately and efficiently reflects everybody’s opinion as to what’s true. Howard Marks Institute, Investments & Wealth, Capturing Inefficiencies: The Rare Insight of Howard Marks (April 1, 2016) . Journal of Investment Consulting, Vol. 17, no. 1, 4-10, 2016, Available at SSRN: [https://ssrn.com/abstract=2792274](https://ssrn.com/abstract=2792274) Dedicated to all the staff of Bowdoin College, from the health services folks and the caf baristas to the library staff and landscapers. Bowdoin is a complex machine that runs so smoothly due to the care and positive energy of many committed employees. My time at Bowdoin has been enriched by everyday interactions with this wonderful community. ACKNOWLEDGMENTS I would like to thank the members of my thesis committee for their help in preparation of this work – Karen Jung, who personally sat down to inform all honors students the submission procedures and guidelines, Carmen Greenlee, for her work organizing copyright guidance, Meredith McCarroll, for informing the requested writing rhetoric and presentation of project. This project would not have been possible without the tremendous support and guidance from Professor David Byrd. As we met throughout the year and he not only patiently answered my questions but also pushed me to think deeply about this topic of research from many different angles. Special thanks are due to the friends and colleagues who made this work possible. Josh-Pablo Patel, Tara Mullen, and J.T. Wooley were all invaluable as both friends and classmates, helping me find time for my work on this project by joining me as group matesand lab partners for the rest of my course load. I am grateful to my partner Jenny for her constant support and willingness to listen to my neural network ramblings at all hours of the day. Finally I would like to thank the facilities of Bowdoin, without Bowdoin’s Slurm HPC Cluster this project would not have been possible for such a short time frame. TABLE OF CONTENTS Acknowledgments ................................... iv List of Tables ...................................... vii List of Figures ..................................... viii List of Acronyms .................................... ix Chapter 1: Introduction and Related Work ..................... 1 1.1 Research Question ............................... 1 1.2 Introduction .................................. 1 1.3 Background and Related work ........................ 2 Chapter 2: Methodology ............................... 5 2.1 Data Sourcing and Collection ......................... 5 2.2 Data Prepossessing ...........

FAQs

What research question does the paper address?

The paper includes a dedicated section for the research question in Chapter 1, framing the study around evaluating sector-based market efficiency.

How is the dataset obtained and prepared?

The methodology describes data sourcing and collection, then data preprocessing before modeling, including preparation of limit order book data samples.

Which models and evaluation measures are used to assess efficiency?

The results section compares model performances, tuning hyper-parameters and reporting sector precision. Adjusted predictive precision is used as a proxy for efficiency, supported by multiple figures and interpretations.

Aug 4, 2026

Upload by Anda