Profiling Investor Behavior in the Malaysian Derivatives Market Using K-Means Clustering

Eng Hao Louis Tan¹Yaman Hamed^1*Hanita Daud¹Mohd Amirul Faiz Abdul Wahab²Ahmad Amirul Adlan Azhar²Sieow Yeek Tan²

• ¹University of Technology Petronas, Tronoh, Malaysia
• ²Bursa Malaysia Berhad, Federal Territory of Kuala Lumpur, Malaysia

This study investigates the trading behaviors of Malaysian derivatives traders using a comprehensive dataset from Bursa Malaysia with K-means clustering, representing one of the first AI applications to derivatives market segmentation. The analysis encompassed over 11 million trade records for FCPO and FKLI derivatives from January to December 2022. Six key features were engineered to segment derivative traders: Total Number of Trades, Total Traded Amount, Overall Realized Profit, Average ROI, Maximum Account Vintage (trader experience in years), and Median Holding Days (typical position duration). Inverse Hyperbolic Sine transformation was applied to address extreme outliers, ensuring robust feature scaling. K-means clustering identified five distinct profiles: "High-Frequency, High-Risk Derivative Traders with Consistent Losses", "Conservative, Steady-Growth Derivative Trader", "High-Frequency, High-Yield Derivative Traders", "Conservative, Low-Yield Derivative Traders", and "Cautious, Low-Activity Novice Derivative Traders". Decision tree classifiers validated these clusters through interpretable splitting conditions. These profiles enable targeted risk management strategies, personalized trading services, and evidence-based regulatory policies for derivatives markets and future research.