Effect of Machine Learning in Better Portfolio Management: Evidence From PSX
Abstract/Description
Portfolio optimization for better profit generation is the desire of every investor. This study attempted to construct fully optimized diverse portfolio with the objective of maximum expected return and best sharpe ratio, after analyzing each stock fundamentally and technically by using machine learning algorithms for prediction and genetic algorithm for optimization. Random Forest (RF) is used for stock return prediction. Data of 170 companies is included in research from all sectors of Pakistan Stock Exchange (PSX) for the time period of 2000 till April 2022. The tests are conducted in three phases. The whole process is based on 4 layers which are analysis layer, prediction layer, diversification layer and optimization layer. Optimization layer uses genetic algorithm for portfolio optimization. Findings of the analysis concluded that the machine learning effects the portfolio optimization positively and it can be used in Pakistan’s market. The predictions of the returns were significant and portfolios were well diversified. Genetic algorithm optimized the portfolio efficiently. The result of the study is significant. This method can be used by portfolio managers for better diversified portfolio construction and its optimization that will help them to maximize the return on portfolio. The methodology also predicts the stock return which is also useful for managers to foresee the future market
Keywords
Portfolio Optimization, Machine Learning, Stock Price Prediction, Random Forest, Genetic Algorithm
Track
Finance
Session Number/Theme
3B: Finance
Session Chair
Dr. Saqib Sharif; Dr. Nauman J. Amin
Start Date/Time
27-5-2023 2:30 PM
End Date/Time
27-5-2023 4:30 PM
Location
MCS-4, AMAN-CED, First Floor
Recommended Citation
Siddiqui, M., & Shah, S. (2023). Effect of Machine Learning in Better Portfolio Management: Evidence From PSX. 3rd IBA SBS International Conference 2024. Retrieved from https://ir.iba.edu.pk/sbsic/2023/program/44
COinS
Effect of Machine Learning in Better Portfolio Management: Evidence From PSX
MCS-4, AMAN-CED, First Floor
Portfolio optimization for better profit generation is the desire of every investor. This study attempted to construct fully optimized diverse portfolio with the objective of maximum expected return and best sharpe ratio, after analyzing each stock fundamentally and technically by using machine learning algorithms for prediction and genetic algorithm for optimization. Random Forest (RF) is used for stock return prediction. Data of 170 companies is included in research from all sectors of Pakistan Stock Exchange (PSX) for the time period of 2000 till April 2022. The tests are conducted in three phases. The whole process is based on 4 layers which are analysis layer, prediction layer, diversification layer and optimization layer. Optimization layer uses genetic algorithm for portfolio optimization. Findings of the analysis concluded that the machine learning effects the portfolio optimization positively and it can be used in Pakistan’s market. The predictions of the returns were significant and portfolios were well diversified. Genetic algorithm optimized the portfolio efficiently. The result of the study is significant. This method can be used by portfolio managers for better diversified portfolio construction and its optimization that will help them to maximize the return on portfolio. The methodology also predicts the stock return which is also useful for managers to foresee the future market
