Modelling Stock Market Manipulation in Online Forums

Loading...
Thumbnail Image

Authors

Nam, David

Date

Type

thesis

Language

eng

Keyword

Market Manipulation , Machine Learning , Online Forums

Research Projects

Organizational Units

Journal Issue

Alternative Title

Abstract

Over the past several decades, advances in technology have significantly impacted all aspects of the financial system. While it has led to numerous benefits, it has also increased the methods for manipulating the market. A frequent platform used to perform these market manipulation schemes has been through social media. In particular, online forums have become a tool for manipulators to disseminate false or misleading information so that they can profit from other investors. As a result, my research provides investors with valuable insights and the tools necessary for detecting pump-and-dump schemes. To achieve this, posts and comments within financial forums were first collected. Then, financial data was added to associate the texts with resulting market behaviours. By using statistical methods, the records were then initially labelled depending on whether they exhibited a known market pattern that commonly occurs when investors act upon deceptive content. To further improve upon the labelling method, comments of deceptive posts were then relabelled based on their level of agreement to fraudulent information. With the described agreement model, results showed that predictions among the tested classification techniques (XGBoost, Random Forest, SVM, MLP, CNN, BiLSTM) were improved. Additionally, by comparing the performance of the classifiers, CNNs were found to be the best performing model among those that were tested.

Description

Citation

Publisher

License

Queen's University's Thesis/Dissertation Non-Exclusive License for Deposit to QSpace and Library and Archives Canada
ProQuest PhD and Master's Theses International Dissemination Agreement
Intellectual Property Guidelines at Queen's University
Copying and Preserving Your Thesis
This publication is made available by the authority of the copyright owner solely for the purpose of private study and research and may not be copied or reproduced except as permitted by the copyright laws without written authority from the copyright owner.

Journal

Volume

Issue

PubMed ID

External DOI

ISSN

EISSN