Forecasting crude oil price (revisited)

摘要: The recent changes in crude oil price behaviour between 2007 and 2009 revived the question about underlying dynamics governing prices. Even more importantly, outstanding over whether we can forecast returns or not needs to be readdressed. goal of this paper is present an analysis spot daily price/returns. aim find if structural market have had effect on ability returns. Also, argue that there still a gap computational methods traditional statistical for time series forecasting; hence, try make effort give due consideration properties building process softcomputing models. As such, our investigation starts with testing non-linearity structure these using most trusted test iid, BDS test. Fuzzy Classifier System (FCS) proposed by Kaboudan (1999) time-domain introduced Barnett Wolff (2005) are also used. Finally, estimate Lyapunov exponents investigate existence chaos return. Our tests show consistently dynamical forces driving non-linear ones, possibly low dimension. Moreover, FCS shows evidence high level noise which means smoothing reduction necessary achieving any accuracy. To short-term compared performance ARIMA(p,d,q), EGARCH(p,q) ANN We conclude it possible models providing control measures results some effective multi-step forecasting (up 26 steps) smoothed