Título: Evolving Functional Fuzzy Model For Interest Rate Term Structure Forecasting
Autores: Maciel, Leandro; Gomide, Fernando; Ballini, Rosangela
Resumo: Evolving fuzzy systems use data streams to continuously adapt the structure and functionality of fuzzy rule-based models. It gradually develops the model structure and its parameters from a stream of data, which is essential when dealing with complex amd nonstationary systems. In this paper, we suggest the use of functional evolving fuzzy modeling in the form of Takagi-Sugeno (eTS) model to forecast Brazilian government bond yields through the Nelson-Siegel function. In this case the eTS adaptively estimates the parameters Nelson-Siegel function to perform forecasts. This is a crucial procedure for bond portfolio management, derivatives and bonds pricing. The experiments reported here use daily data of the Brazilian National Treasury Bills of the period from January 2007 to December 2009 for one, three, six, nine and twelve months ahead forecasting horizons. The evolving model was compared with autoregressive and random walk models, in terms of root mean squared error. Results indicate that eTS is a promising approach to deal with government bond yields forecasting because it gives more accurate Nelson-Siegel parameters values than traditional approaches.
Palavras-chave: Evolving Fuzzy Systems; Time Series Forecasting; Yield Curve; Interest Rate
Código DOI: 10.21528/CBIC2011-18.3
Artigo em pdf: st_18.3.pdf
Arquivo BibTex: st_18.3.bib