Estimação e previsão da estrutura a termo das taxas de juros usando técnicas de inteligência computacional / Term structure of interest rate modeling and forecasting using computational intelligence techniques
AUTOR(ES)
Leandro dos Santos Maciel
FONTE
IBICT - Instituto Brasileiro de Informação em Ciência e Tecnologia
DATA DE PUBLICAÇÃO
25/06/2012
RESUMO
This work proposes the term structure of interest rates modeling and forecasting using computational intelligence techniques, based on data from the US and Brazilian fixed income markets. The yield curve modeling includes the use of some evolutionary computation methods like Genetic Algorithms, Differential Evolution and Evolution Strategies in comparison with traditional optimization techniques such as nonlinear least squares and sequential quadratic programming. The motivation behind the use of evolutionary computation to yield curve estimation aims to overcome limitations like non-convergence and high parameters instability to initialization. Moreover, recently, the literature has been shown the higher performance of genetic algorithms in yield curve modeling problems. This work also contributes by developing an extended participatory learning fuzzy model, called ePL+, which includes on its original version, ePL, mechanisms to improve its autonomy and adaptability in complex systems modeling. Therefore, the ePL+ model and some evolving functional fuzzy approaches were evaluated in the future interest rates forecasting, as opposed to econometric models based on autoregressive processes and multilayer artificial neural networks methodologies, since interest rates evolution shows a high non-linear dynamics and also time-varying, justifying the idea of adaptive modeling. Models performance were compared in terms of error measures, computational complexity and by parametric and non-parametric statistical tests, MGN and SIGN, respectively. The results reveal the high potential of computational intelligence methods to deal with the term structure modeling and forecasting for both economies considered, as pointed out by their adjustment and statistical superior performance then traditional optimization and econometrics techniques reported in the literature.
ASSUNTO(S)
taxas de juros computação evolutiva sistemas nebulosos sistemas de computação adaptativos análise de séries temporais interest rates evolutionary computation fuzzy systems adaptive computing systems time-series analysis
ACESSO AO ARTIGO
http://libdigi.unicamp.br/document/?code=000869281Documentos Relacionados
- IDENTIFICAÇÃO DE MODELOS GARCH USANDO INTELIGÊNCIA COMPUTACIONAL
- PREVISÃO DA CARGA DE ENERGIA DIÁRIA EM PATAMARES COMBINANDO TÉCNICAS ESTATÍSTICAS E DE INTELIGÊNCIA COMPUTACIONAL
- Teoria da estrutura a termo das taxas de juros
- Dois ensaios em econometria financeira: modelagem, previsão e stress da estrutura a termo das taxas de juros
- UM MODELO DE PREVISÃO DE CURTO PRAZO DE CARGA ELÉTRICA COMBINANDO MÉTODOS ESTATÍSTICOS E INTELIGÊNCIA COMPUTACIONAL