Portfolio Selection Based on Factorial Heteroskedastic Models: Application to Fund of Funds

• Autor(a) principal: Caldeira, Joao
• Data de Publicação: 2014
• Outros Autores: Moura, Guilherme, Portela, André, Tessari, Cristina
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: RAM. Revista de Administração Mackenzie
• Texto Completo: https://editorarevistas.mackenzie.br/index.php/RAM/article/view/4831
• topic: Multivariate GARCH. Dynamic Conditional Correlation (DCC). Performance Evaluation. Fund of Funds. Portfolio Optimization.; GARCH Multivariado. Correlación Condicional Dinámica (DCC). Evaluación del Desempeño. Fondo de Fondos. Optimización de Carteras.; GARCH multivariado; correlação condicional dinâmica – DCC; análise de desempenho; fundo de fundos; otimização de carteiras.

The modern portfolio theory is based on the idea that diversification of a portfolio results in a better relationship between risk and return. Recently, managers have tried to extend the diversification of their portfolios by investing in fund of funds that, in turn, already contains diversified portfolios. With that comes growing interest from academic and market participants in the selection of portfolios formed by investment funds. In this paper, the applicability and performance out of sample of quantitative portfolio optimization strategies to build portfolios of funds will be analyzed. The performance of these portfolios of investment funds will be compared with the performance of the naive equally weighted portfolio, the Ibovespa index and the fixed income index, IRF-M. To obtain optimal portfolios, restricted to short selling, we determine an optimization problem of portfolios composed of 388 Brazilian hedge (multimarket) funds over five years traded in the Brazilian market. For modeling of the covariance matrix of returns of 388 funds is used a heteroscedastic factorial parsimonious model. Considering different frequencies for the portfolio re-balancing frequency (daily, weekly and monthly), the measures of performance out of the sample show that the optimal portfolios exhibit superior results in terms of volatility, risk-adjusted performance, turnover and transaction costs over time. In particular, the results based on robust statistical tests indicate that the Sharpe ratio (SR) of the mean-variance portfolios and that of the minimum-variance portfolios were statistically different (higher) compared to the SR of the benchmark index in all portfolio re-balancing frequencies used in the paper. Regarding the standard deviation, statistical tests employed in the paper showed that the volatility of the minimum-variance portfolios is statistically different (lower) than the volatility of the benchmark index. Similar results were found when comparing the performance of the optimal portfolios with respect to the Ibovespa index and the equally- weighted portfolio.