Publicação
Machine learning applications in portfolio management theory
| Resumo: | Portfolio management, being the practice of managing and selecting an investment strategy and allocation for a defined investor, has always aimed at maximizing return while minimizing the risk of a combination of financial securities, hence a portfolio. The financial world has been evolving since Markovitz introduced the modern portfolio theory (MPT) in 1952, although nowadays it is still widely addressed as the benchmark and foundation for optimization methods. Traditional techniques of portfolio allocation such as MPT were considered without flaws for many decades, however its implications and notions were utilized to create enhanced several newer theories over the years, such as capital asset pricing theory (CAPM), arbitrage pricing theory (APT) and many others. The technologic advancement introduced computing power and Artificial Intelligence (AI) techniques into the industry, creating the possibility of handling large and complex datasets through instructed algorithms. The scope of this analysis was to employ the oldest and most popular approach such as MPT in combination with the Monte-Carlo method, a stochastic model to simulate random portfolio, and create an investment strategy based on these assumptions. Machine Learning (ML) models were then applied to analyse their impact on the previous strategy. Specifically, a clustering algorithm was implemented to reach a high level of diversification, while an auto-regression model, such as ARIMA, aimed at predicting future stock prices. The project utilized historical data to compute the analysis and each strategy was back-tested over four years to evaluate their accuracy and performance and compared with a benchmark index, Standards and Poor (S&P 500) in this case. The results of the machine learning-based techniques showed a higher performance compared to the index benchmark, indicating a well-diversified portfolio due to the clustering algorithm and an acceptable level of accuracy for the ARIMA model. The portfolio randomly constructed displayed the lowest performance out of all the strategies and the benchmark index, since the stocks selection did not provide a high degree of diversification. |
|---|---|
| Autores principais: | Neri, Marco |
| Assunto: | MPT diversification Machine Learning Monte-Carlo ARIMA diversificação |
| Ano: | 2023 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório da Universidade de Lisboa |
| _version_ | 1866811010524905472 |
|---|---|
| author | Neri, Marco |
| author_facet | Neri, Marco |
| author_role | author |
| contributor_name_str_mv | Vieira, Pedro Rino Repositório Científico de Acesso Aberto da ULisboa |
| country_str | PT |
| creators_json_txt | [{\"Person.name\":\"Neri, Marco\"}] |
| datacite.contributors.contributor.contributorName.fl_str_mv | Vieira, Pedro Rino Repositório Científico de Acesso Aberto da ULisboa |
| datacite.creators.creator.creatorName.fl_str_mv | Neri, Marco |
| datacite.date.Accepted.fl_str_mv | 2023-07-01T00:00:00Z |
| datacite.date.available.fl_str_mv | 2024-04-13T00:30:31Z |
| datacite.date.embargoed.fl_str_mv | 2024-04-13T00:30:31Z |
| datacite.rights.fl_str_mv | http://purl.org/coar/access_right/c_abf2 |
| datacite.subjects.subject.fl_str_mv | MPT diversification Machine Learning Monte-Carlo ARIMA diversificação |
| datacite.titles.title.fl_str_mv | Machine learning applications in portfolio management theory |
| dc.contributor.none.fl_str_mv | Vieira, Pedro Rino Repositório Científico de Acesso Aberto da ULisboa |
| dc.creator.none.fl_str_mv | Neri, Marco |
| dc.date.Accepted.fl_str_mv | 2023-07-01T00:00:00Z |
| dc.date.available.fl_str_mv | 2024-04-13T00:30:31Z |
| dc.date.embargoed.fl_str_mv | 2024-04-13T00:30:31Z |
| dc.format.none.fl_str_mv | application/pdf |
| dc.identifier.none.fl_str_mv | http://hdl.handle.net/10400.5/29020 |
| dc.language.none.fl_str_mv | eng |
| dc.publisher.none.fl_str_mv | Instituto Superior de Economia e Gestão |
| dc.rights.none.fl_str_mv | http://purl.org/coar/access_right/c_abf2 |
| dc.subject.none.fl_str_mv | MPT diversification Machine Learning Monte-Carlo ARIMA diversificação |
| dc.title.fl_str_mv | Machine learning applications in portfolio management theory |
| dc.type.none.fl_str_mv | http://purl.org/coar/resource_type/c_bdcc |
| description | Portfolio management, being the practice of managing and selecting an investment strategy and allocation for a defined investor, has always aimed at maximizing return while minimizing the risk of a combination of financial securities, hence a portfolio. The financial world has been evolving since Markovitz introduced the modern portfolio theory (MPT) in 1952, although nowadays it is still widely addressed as the benchmark and foundation for optimization methods. Traditional techniques of portfolio allocation such as MPT were considered without flaws for many decades, however its implications and notions were utilized to create enhanced several newer theories over the years, such as capital asset pricing theory (CAPM), arbitrage pricing theory (APT) and many others. The technologic advancement introduced computing power and Artificial Intelligence (AI) techniques into the industry, creating the possibility of handling large and complex datasets through instructed algorithms. The scope of this analysis was to employ the oldest and most popular approach such as MPT in combination with the Monte-Carlo method, a stochastic model to simulate random portfolio, and create an investment strategy based on these assumptions. Machine Learning (ML) models were then applied to analyse their impact on the previous strategy. Specifically, a clustering algorithm was implemented to reach a high level of diversification, while an auto-regression model, such as ARIMA, aimed at predicting future stock prices. The project utilized historical data to compute the analysis and each strategy was back-tested over four years to evaluate their accuracy and performance and compared with a benchmark index, Standards and Poor (S&P 500) in this case. The results of the machine learning-based techniques showed a higher performance compared to the index benchmark, indicating a well-diversified portfolio due to the clustering algorithm and an acceptable level of accuracy for the ARIMA model. The portfolio randomly constructed displayed the lowest performance out of all the strategies and the benchmark index, since the stocks selection did not provide a high degree of diversification. |
| dirty | 0 |
| eu_rights_str_mv | openAccess |
| format | masterThesis |
| fulltext.url.fl_str_mv | https://repositorio.ulisboa.pt/bitstreams/5eb571a4-6afe-43ad-9514-3e2c0b32ffa7/download |
| id | ul_a342c6526ca8d2fd64e2012b701ec11d |
| identifier.url.fl_str_mv | http://hdl.handle.net/10400.5/29020 |
| instacron_str | ul |
| institution | Universidade de Lisboa |
| instname_str | Universidade de Lisboa |
| language | eng |
| network_acronym_str | ul |
| network_name_str | Repositório da Universidade de Lisboa |
| oai_identifier_str | oai:repositorio.ulisboa.pt:10400.5/29020 |
| organization_str_mv | urn:organizationAcronym:ul |
| person_str_mv | Neri, Marco |
| publishDate | 2023 |
| publisher.none.fl_str_mv | Instituto Superior de Economia e Gestão |
| reponame_str | Repositório da Universidade de Lisboa |
| repository_id_str | urn:repositoryAcronym:ul |
| service_str_mv | urn:repositoryAcronym:ul |
| spelling | engInstituto Superior de Economia e Gestãopt_PTPortfolio management, being the practice of managing and selecting an investment strategy and allocation for a defined investor, has always aimed at maximizing return while minimizing the risk of a combination of financial securities, hence a portfolio. The financial world has been evolving since Markovitz introduced the modern portfolio theory (MPT) in 1952, although nowadays it is still widely addressed as the benchmark and foundation for optimization methods. Traditional techniques of portfolio allocation such as MPT were considered without flaws for many decades, however its implications and notions were utilized to create enhanced several newer theories over the years, such as capital asset pricing theory (CAPM), arbitrage pricing theory (APT) and many others. The technologic advancement introduced computing power and Artificial Intelligence (AI) techniques into the industry, creating the possibility of handling large and complex datasets through instructed algorithms. The scope of this analysis was to employ the oldest and most popular approach such as MPT in combination with the Monte-Carlo method, a stochastic model to simulate random portfolio, and create an investment strategy based on these assumptions. Machine Learning (ML) models were then applied to analyse their impact on the previous strategy. Specifically, a clustering algorithm was implemented to reach a high level of diversification, while an auto-regression model, such as ARIMA, aimed at predicting future stock prices. The project utilized historical data to compute the analysis and each strategy was back-tested over four years to evaluate their accuracy and performance and compared with a benchmark index, Standards and Poor (S&P 500) in this case. The results of the machine learning-based techniques showed a higher performance compared to the index benchmark, indicating a well-diversified portfolio due to the clustering algorithm and an acceptable level of accuracy for the ARIMA model. The portfolio randomly constructed displayed the lowest performance out of all the strategies and the benchmark index, since the stocks selection did not provide a high degree of diversification.application/pdfpt_PTMachine learning applications in portfolio management theoryNeri, MarcoVieira, Pedro RinoHostingInstitutionOrganizationalRepositório Científico de Acesso Aberto da ULisboae-mailmailto:repositorio@reitoria.ulisboa.ptrepositorio@reitoria.ulisboa.pt2024-04-13T00:30:31Z2023-072023-07-01T00:00:00ZHandlehttp://hdl.handle.net/10400.5/29020http://purl.org/coar/access_right/c_abf2open accessMPTdiversificationMachine LearningMonte-CarloARIMAdiversificação3917319 bytesliteraturehttp://purl.org/coar/resource_type/c_bdccmaster thesishttp://purl.org/coar/access_right/c_abf2application/pdffulltexthttps://repositorio.ulisboa.pt/bitstreams/5eb571a4-6afe-43ad-9514-3e2c0b32ffa7/download |
| spellingShingle | Machine learning applications in portfolio management theory Neri, Marco MPT diversification Machine Learning Monte-Carlo ARIMA diversificação |
| status | SINGLETON |
| subject.fl_str_mv | MPT diversification Machine Learning Monte-Carlo ARIMA diversificação |
| title | Machine learning applications in portfolio management theory |
| title_full | Machine learning applications in portfolio management theory |
| title_fullStr | Machine learning applications in portfolio management theory |
| title_full_unstemmed | Machine learning applications in portfolio management theory |
| title_short | Machine learning applications in portfolio management theory |
| title_sort | Machine learning applications in portfolio management theory |
| topic | MPT diversification Machine Learning Monte-Carlo ARIMA diversificação |
| topic_facet | MPT diversification Machine Learning Monte-Carlo ARIMA diversificação |
| url | http://hdl.handle.net/10400.5/29020 |
| visible | 1 |