Performance profiling of the unit trust funds in Malaysia with data mining techniques

Area of Science:

• Data mining applications in finance
• Investment analytics for retail investors

Background:

• Millennials seek investment income, with unit trusts being a popular option.
• Analyzing unit trust financial data is complex for individuals lacking financial expertise or time.
• Compiling performance data for Malaysian unit trusts presents challenges.

Purpose of the Study:

• To identify unit trust funds offering superior returns compared to their peers through performance profiling.
• To assist amateur retail investors in selecting appropriate unit trusts based on risk tolerance.

Main Methods:

• Employed Expectation Maximization (EM) for clustering Malaysian unit trust funds by annual financial performance.
• Utilized Apriori algorithm to derive association rules for profiling clustered unit trust funds.
• Performance profiling based on yearly average total returns for 2018 and 2019.

Main Results:

• Clustering successfully grouped unit trust funds based on financial performance.
• Derived association rules provided valuable insights for profiling fund clusters.
• Evaluation confirmed the utility of profiling for investors with diverse risk appetites.

Conclusions:

• Data mining approaches can effectively achieve financial performance profiling of unit trust funds.
• Performance profiling offers crucial information for unit trust investors to select suitable investment funds.