| Abstract: | Market manipulation poses a significant threat to the integrity and efficiency of financial markets
and commodity markets, particularly in emerging markets such as the Ethiopian Commodity
Exchange (ECX). This thesis aims to develop a robust machine-learning model to detect and
mitigate market manipulation within the ECX. By leveraging historical transaction data and
employing advanced machine learning algorithms, the study seeks to identify anomalous trading
patterns indicative of manipulative activities.
The research begins with a comprehensive review of the existing literature on market manipulation
detection and machine learning techniques. Subsequently, a detailed analysis of the ECX's trading
data is conducted to understand the unique characteristics and potential vulnerabilities of this
market. Data preprocessing techniques are employed to cleanse and prepare the data for model
training.
Various machine learning models, including supervised and unsupervised learning algorithms,
are evaluated for their efficacy in detecting market manipulation. The models are trained on
labeled datasets containing instances of known manipulative activities and normal trading
behavior. Performance metrics such as accuracy, precision, recall, and F1-score are used to
assess the effectiveness of each model.
The results demonstrate that certain machine learning models, particularly ensemble methods and
neural networks, show high potential in accurately detecting market manipulation within the ECX.
The best-performing model is integrated into a real-time monitoring system, providing timely
alerts to market regulators and stakeholders.
This study contributes to the body of knowledge by offering a novel approach to market
manipulation detection in commodity exchanges, with a specific focus on the Ethiopian context.
The developed model not only enhances market surveillance capabilities but also promotes market
integrity, investor confidence, and overall market stability.
Future work will explore the scalability of the model to other emerging markets and the
incorporation of additional data sources, such as social media sentiment and economic indicators,
to further enhance the model's predictive power. |
