| Author: | Cui, Yi |
| Title: | Machine learning for financial analysts |
| Advisors: | Wei, Xiangdong (AF) Guo, Song (COMP) |
| Degree: | DFinTech |
| Year: | 2022 |
| Subject: | Finance -- Data processing Machine learning Hong Kong Polytechnic University -- Dissertations |
| Department: | Faculty of Business |
| Pages: | ii, 196 pages : color illustrations |
| Language: | English |
| Abstract: | Finance industry is embracing technology. The use of machine learning (ML) methods for economic research is popular, though it is considered to be a “black box.” This research proposes an ML-based automatic stock selection method that relies on the fundamental analysis approach, and the key is the development of an explainable artificial intelligence (AI) model. Data used for this research are modeled based on A-listed shares data with the China Stock Market & Accounting Research (CSMAR) database. To further explore the potential of this ML method, U.S. stock data from Compustat are also regressed in the research with the exact same methodology, and they produce similarly good results. Data evidence in this research have exhibited the “unbox” process of ML-based fundamental analysis. The theory behind this ML research is based on financial statement analysis, which dates back to the 1960s. The attributes and distribution of financial statement data are carefully examined in this study using computer-based tools, because ML has significantly expanded the breadth and depth of data usage. Machine is capable of reading through more than 30,000 financial statements at once and memorize them all. Financial statement data is processed in highest efficiency than ever. Two methods are introduced to enhance the accuracy of prediction. 1) Be voracious regarding data, instead of simplifying. Contrary to traditional financial research, ML does not need a simple data model; instead, it requires redundancy to secure the model’s consistency without adding burden to its computing capability. 127 accounts are at least imported for final model. 2) Separate data into positives and negatives. Contrary to traditional accounting knowledge, negative numbers in accounts are more frequent than normal. This separation handling significantly improves prediction accuracy. The output of this research aims to provide 1) insightful predictive data for a specific stock, 2) visual and statistical evidence of its prediction, 3) and a “usable-to-all” solution. In this research, fundamental analysis is used as the cornerstone of stock price prediction, which leaves room for the discussion of investment strategies. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 6732.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 11.98 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/12213