Using a simple unobserved components model, we show that explicitly modelling asymmetric cycles on crude oil prices improves the forecast ability of univariate time series models of the oil price.