Sporala red del conocimiento
EN
Universidad EAFIT

Maestría en Administración Financiera · 2024

Evaluación del efecto de incluir la predicción de rendimientos mediante la técnica de Support Vector Machines en la eficiencia del modelo de media-varianza de Markowitz

Portfolio investment optimization aims to maximize expected returns given certain levels of risk. This process requires dealing with different variables in a nonlinear, noisy system due to market complexity. This is understood as a system that is affected by different external conditions that may be uncontrollable, where volatility influenced by unpredictable factors is present. In this study, an analysis of the results obtained by integrating machine learning techniques, specifically the set of algorithms called Support Vector Machines (SVM), into classical portfolio construction models is conducted. These algorithms allow for the analysis of large amounts of data and the estimation of asset return time series, resulting in a hybrid optimization model. Historical data from the stock markets of the United States and Colombia are used for numerical experiments; one set of data is used for model training (Training Set) and another for testing (Testing Set). Finally, the efficiency of the model is evaluated comparatively with the mean-variance portfolio selection theory proposed by Markowitz.

Texto completo 49 páginas con texto

Leer la tesis completa Ficha en el repositorio

Contenido

Palabras clave

Tesauro de su biblioteca