Thesis REPLICACIÓN DE ÍNDICES MEDIANTE ALGORITMOS GENÉTICOS
Loading...
Date
2017
Journal Title
Journal ISSN
Volume Title
Program
INGENIERIA CIVIL INDUSTRIAL
Campus
Universidad Técnica Federico Santa María UTFSM. Casa Central Valparaíso
Abstract
En este trabajo se analiza el uso de algoritmos genéticos para la generación de portafolios que asemejen los retornos de un índice bursátil bajo una serie de restricciones que generalmente enfrentan los administradores de fondos de inversión pasiva.Precios de cierre semanal para cinco índices bursátiles son utilizados. En una primera instancia se contrastan los resultados obtenidos mediante el uso de algoritmos genéticos con aquellos obtenidos mediante métodos de programación matemático para problemas de menor tamaño, obteniendo resultados muy cercanos entre ambos métodos. Posteriormente, se analiza el comportamiento para índices de mayor tamaño, donde el número de variables involucradas imposibilita la resolución mediante métodos tradicionales, obteniendo buenos resultados en tan solo unos minutos. Finalmente, se analiza un segundo operador de mutación diseñado con el fin de evitar la convergencia prematura en óptimos locales.
In this paper, we analyze the use of Genetic algorithms as a technic to generate an investment portfolio capable of replicate the returns of an index under a series of constrains which passive portfolio managers usually face.Weekly closing prices of five major stock market indices are used. Initially, we compare the results of the portfolios provided by the genetic algorithm against the ones from mathematical programming for a small size problem. After that, we analyze the behavior of the algorithm against larger indices, on which the number of variables limits the resolution by means of traditional optimization methods. Lastly, a new mutation operator is proposed, which improves the efficiency of the algorithm avoiding this from getting trapped in local optimums.
In this paper, we analyze the use of Genetic algorithms as a technic to generate an investment portfolio capable of replicate the returns of an index under a series of constrains which passive portfolio managers usually face.Weekly closing prices of five major stock market indices are used. Initially, we compare the results of the portfolios provided by the genetic algorithm against the ones from mathematical programming for a small size problem. After that, we analyze the behavior of the algorithm against larger indices, on which the number of variables limits the resolution by means of traditional optimization methods. Lastly, a new mutation operator is proposed, which improves the efficiency of the algorithm avoiding this from getting trapped in local optimums.
Description
Catalogado desde la version PDF de la tesis.
Keywords
ADMINISTRACION DE PORTAFOLIOS, ALGORITMOS GENETICOS, HEURISTICAS, REPLICACION DE INDICES