Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning

In the current context, the stock market has become a complex and ever-changing industry, where decision-making can be crucial for making significant profits or incurring significant losses. The evolution of investment systems has led to the incorporation of innovative machine learning techniques, s...

Descripción completa

Detalles Bibliográficos
Autor: Such Ballester, Ignacio
Tipo de recurso: tesis de maestría
Fecha de publicación:2024
País:España
Institución:Universitat Oberta de Catalunya (UOC)
Repositorio:O2, repositorio institucional de la UOC
OAI Identifier:oai:openaccess.uoc.edu:10609/149728
Acceso en línea:http://hdl.handle.net/10609/149728
Access Level:acceso abierto
Palabra clave:deep reinforcement learning
stock market
proximal policy optimization
aprendizaje por refuerzo
mercado de valores
optimización de políticas próximas
Reinforcement learning -- TFM
Aprenentatge per reforç -- TFM
id ES_7c559710d1912b43c8dbd85b51eabc24
oai_identifier_str oai:openaccess.uoc.edu:10609/149728
network_acronym_str ES
network_name_str España
repository_id_str
spelling Agente inversor para acciones de small cap mediante el uso de Reinforcement LearningSuch Ballester, Ignaciodeep reinforcement learningstock marketproximal policy optimizationaprendizaje por refuerzomercado de valoresoptimización de políticas próximasReinforcement learning -- TFMAprenentatge per reforç -- TFMIn the current context, the stock market has become a complex and ever-changing industry, where decision-making can be crucial for making significant profits or incurring significant losses. The evolution of investment systems has led to the incorporation of innovative machine learning techniques, such as Reinforcement Learning, which allow them to learn and adapt to market changes in real-time. Reinforcement Learning is a branch of machine learning based on the concept of reward and punishment, which aims to maximise the reward obtained through interaction with an environment. This technique has proven its effectiveness in solving complex problems and has been successfully applied in environments such as robotics and video games. In this context, the use of Reinforcement Learning in the stock market presents itself as a promising alternative for the design of optimal and profitable investment strategies in a changing and highly competitive environment. The objective of this Master’s thesis is to establish a new line of research in predicting stock values of ”small cap¸companies through the use of Deep Reinforcement Learning algorithms. The first algorithm is the Proximal Policy Optimization (PPO), where the implementation of Liu et al. [11] will be used. On the other hand, the A2C and DDPG algorithms will be employed, which have been promising according to this paper Liu et al.En el contexto actual, el mercado bursátil se ha convertido en una industria compleja y en constante cambio, donde la toma de decisiones puede ser crucial para obtener beneficios o incurrir en pérdidas significativas. La evolución de los sistemas de inversión ha llevado a la incorporación de técnicas innovadoras de aprendizaje automático, como el Reinforcement Learning, que permiten aprender y adaptarse a los cambios del mercado en tiempo real. El Reinforcement Learning es una rama del aprendizaje automático que se basa en el concepto de premio y castigo, y su objetivo es maximizar la recompensa obtenida a través de un proceso de interacción con un entorno. Esta técnica ha demostrado su eficacia en la resolución de problemas complejos y ha sido aplicada con éxito en entornos como la robótica y los videojuegos. En este contexto, el uso del Reinforcement Learning en el mercado bursátil se presenta como una alternativa prometedora para el diseño de estrategias de inversión óptimas y rentables en un entorno cambiante y altamente competitivo. El objetivo de esta tesis de fin de Máster es establecer una nueva línea de investigación en la predicción de valores de las acciones de empresas ”small cap” mediante el uso de algoritmos de Deep Reinforcement Learning. El primer algoritmo es el Proximal Policy Optimization (PPO) A, donde se utilizará la implementación de Liu et al. [11]. En ella, se muestra como es un buen agente de bolsa en momentos al alza pero son mas vulnerables en etapas de descenso. Por otro lado, se mencionan los algoritmos A2C C y B, los cuales han sido prometedores en la misma obra. Se demuestra que el DDPG, no es tan eficaz como el PPO, aunque sí es mas cauto en etapas de caídas en bolsa.Universitat Oberta de Catalunya (UOC)Benito Altamirano, IsmaelPérez Ibáñez, Rubén202420242024info:eu-repo/semantics/masterThesisapplication/pdfapplication/pdfhttp://hdl.handle.net/10609/149728reponame:O2, repositorio institucional de la UOCinstname:Universitat Oberta de Catalunya (UOC)EspañolCC BY-NC-NDhttp://creativecommons.org/licenses/by-nc-nd/3.0/es/info:eu-repo/semantics/openAccessoai:openaccess.uoc.edu:10609/1497282026-05-28T12:42:01Z
dc.title.none.fl_str_mv Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
title Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
spellingShingle Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
Such Ballester, Ignacio
deep reinforcement learning
stock market
proximal policy optimization
aprendizaje por refuerzo
mercado de valores
optimización de políticas próximas
Reinforcement learning -- TFM
Aprenentatge per reforç -- TFM
title_short Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
title_full Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
title_fullStr Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
title_full_unstemmed Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
title_sort Agente inversor para acciones de small cap mediante el uso de Reinforcement Learning
dc.creator.none.fl_str_mv Such Ballester, Ignacio
author Such Ballester, Ignacio
author_facet Such Ballester, Ignacio
author_role author
dc.contributor.none.fl_str_mv Benito Altamirano, Ismael
Pérez Ibáñez, Rubén
dc.subject.none.fl_str_mv deep reinforcement learning
stock market
proximal policy optimization
aprendizaje por refuerzo
mercado de valores
optimización de políticas próximas
Reinforcement learning -- TFM
Aprenentatge per reforç -- TFM
topic deep reinforcement learning
stock market
proximal policy optimization
aprendizaje por refuerzo
mercado de valores
optimización de políticas próximas
Reinforcement learning -- TFM
Aprenentatge per reforç -- TFM
description In the current context, the stock market has become a complex and ever-changing industry, where decision-making can be crucial for making significant profits or incurring significant losses. The evolution of investment systems has led to the incorporation of innovative machine learning techniques, such as Reinforcement Learning, which allow them to learn and adapt to market changes in real-time. Reinforcement Learning is a branch of machine learning based on the concept of reward and punishment, which aims to maximise the reward obtained through interaction with an environment. This technique has proven its effectiveness in solving complex problems and has been successfully applied in environments such as robotics and video games. In this context, the use of Reinforcement Learning in the stock market presents itself as a promising alternative for the design of optimal and profitable investment strategies in a changing and highly competitive environment. The objective of this Master’s thesis is to establish a new line of research in predicting stock values of ”small cap¸companies through the use of Deep Reinforcement Learning algorithms. The first algorithm is the Proximal Policy Optimization (PPO), where the implementation of Liu et al. [11] will be used. On the other hand, the A2C and DDPG algorithms will be employed, which have been promising according to this paper Liu et al.
publishDate 2024
dc.date.none.fl_str_mv 2024
2024
2024
dc.type.none.fl_str_mv info:eu-repo/semantics/masterThesis
format masterThesis
dc.identifier.none.fl_str_mv http://hdl.handle.net/10609/149728
url http://hdl.handle.net/10609/149728
dc.language.none.fl_str_mv Español
language_invalid_str_mv Español
dc.rights.none.fl_str_mv CC BY-NC-ND
http://creativecommons.org/licenses/by-nc-nd/3.0/es/
info:eu-repo/semantics/openAccess
rights_invalid_str_mv CC BY-NC-ND
http://creativecommons.org/licenses/by-nc-nd/3.0/es/
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
application/pdf
dc.publisher.none.fl_str_mv Universitat Oberta de Catalunya (UOC)
publisher.none.fl_str_mv Universitat Oberta de Catalunya (UOC)
dc.source.none.fl_str_mv reponame:O2, repositorio institucional de la UOC
instname:Universitat Oberta de Catalunya (UOC)
instname_str Universitat Oberta de Catalunya (UOC)
reponame_str O2, repositorio institucional de la UOC
collection O2, repositorio institucional de la UOC
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869411582322147329
score 15,30478