Genetic algorithm for portfolio optimization
This paper aims to study the efficiency of introducing variations in the Genetic Algorithm (GA) shown by Sefiane and Benbouziane in “Portfolio Selection using Genetic Algorithm” in order to optimize a multi-objective problem, which in this case is portfolio optimization. There can be multiple soluti...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/64182 |
| _version_ | 1811687678251368448 |
|---|---|
| author | Upadhyay, Sanjana |
| author2 | Wang Libo |
| author_facet | Wang Libo Upadhyay, Sanjana |
| author_sort | Upadhyay, Sanjana |
| collection | NTU |
| description | This paper aims to study the efficiency of introducing variations in the Genetic Algorithm (GA) shown by Sefiane and Benbouziane in “Portfolio Selection using Genetic Algorithm” in order to optimize a multi-objective problem, which in this case is portfolio optimization. There can be multiple solutions to an optimal portfolio of a fixed number of stocks depending on the risk appetite of the investor, which are represented on Markowitz’s Efficient Frontier. Higher the required return, greater will be the risk taken. In this paper, results on GA optimization obtained by Sefiane and Benbouziane are further explored using the same data-set, but by changing genetic operator parameters as well as constraints on the portfolio, drawing from the work of Jeffrey Horn and David Goldberg in "A Niched Pareto Genetic Algorithm for Multiobjective Optimization"as well as that of “M. Srinivas and L.M. Patnaik in “Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms”. In this study, a fitness function allocating equal weightage to both return and risk is defined as part of a genetic algorithm, to obtain the weights of each of the components of the optimal portfolio. The performance of the GA is improved as compared to the paper by Sefiane and Benbouziane by varying the parameters of the two genetic operators used in the algorithm, namely crossover and mutation. It can be clearly observed that choice of fitness function, which is different in our case as compared to previous prominent works, does affect the results obtained from the GA, and can be modeled according to the user’s needs. We see that the GA can be used as a powerful tool to help the investor manage his wealth better, in both cases of constrained as well as unconstrained optimization. |
| first_indexed | 2024-10-01T05:20:08Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/64182 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T05:20:08Z |
| publishDate | 2015 |
| record_format | dspace |
| spelling | ntu-10356/641822023-07-07T17:07:00Z Genetic algorithm for portfolio optimization Upadhyay, Sanjana Wang Libo School of Electrical and Electronic Engineering DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence DRNTU::Business::Finance::Portfolio management DRNTU::Business::Finance::Asset allocation DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems This paper aims to study the efficiency of introducing variations in the Genetic Algorithm (GA) shown by Sefiane and Benbouziane in “Portfolio Selection using Genetic Algorithm” in order to optimize a multi-objective problem, which in this case is portfolio optimization. There can be multiple solutions to an optimal portfolio of a fixed number of stocks depending on the risk appetite of the investor, which are represented on Markowitz’s Efficient Frontier. Higher the required return, greater will be the risk taken. In this paper, results on GA optimization obtained by Sefiane and Benbouziane are further explored using the same data-set, but by changing genetic operator parameters as well as constraints on the portfolio, drawing from the work of Jeffrey Horn and David Goldberg in "A Niched Pareto Genetic Algorithm for Multiobjective Optimization"as well as that of “M. Srinivas and L.M. Patnaik in “Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms”. In this study, a fitness function allocating equal weightage to both return and risk is defined as part of a genetic algorithm, to obtain the weights of each of the components of the optimal portfolio. The performance of the GA is improved as compared to the paper by Sefiane and Benbouziane by varying the parameters of the two genetic operators used in the algorithm, namely crossover and mutation. It can be clearly observed that choice of fitness function, which is different in our case as compared to previous prominent works, does affect the results obtained from the GA, and can be modeled according to the user’s needs. We see that the GA can be used as a powerful tool to help the investor manage his wealth better, in both cases of constrained as well as unconstrained optimization. Bachelor of Engineering 2015-05-25T05:26:28Z 2015-05-25T05:26:28Z 2015 2015 Final Year Project (FYP) http://hdl.handle.net/10356/64182 en Nanyang Technological University 59 p. application/pdf |
| spellingShingle | DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence DRNTU::Business::Finance::Portfolio management DRNTU::Business::Finance::Asset allocation DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems Upadhyay, Sanjana Genetic algorithm for portfolio optimization |
| title | Genetic algorithm for portfolio optimization |
| title_full | Genetic algorithm for portfolio optimization |
| title_fullStr | Genetic algorithm for portfolio optimization |
| title_full_unstemmed | Genetic algorithm for portfolio optimization |
| title_short | Genetic algorithm for portfolio optimization |
| title_sort | genetic algorithm for portfolio optimization |
| topic | DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence DRNTU::Business::Finance::Portfolio management DRNTU::Business::Finance::Asset allocation DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems |
| url | http://hdl.handle.net/10356/64182 |
| work_keys_str_mv | AT upadhyaysanjana geneticalgorithmforportfoliooptimization |