Forecasting WTI crude oil futures returns: Does the term structure help?
研究发现从WTI原油期货期限结构中提取的Nelson-Siegel因子能预测持有期收益，结合宏观指标或LASSO模型可降低预测误差，并提升交易策略的夏普比率和偏度表现。
Nelson-Siegel (NS) factors extracted from the term structure of WTI oil futures are shown to predict subsequent WTI holding period returns in-sample. This in-sample predictability is not diminished by augmenting with macroeconomic indicators or oil market specific predictors. Allowing the decay factor in the Nelson-Siegel model to vary over time improves in-sample predictions at medium horizon return forecasts. We conduct out-of-sample forecasting exercises on models that use NS factors, such as a simple two factor model that uses a composite leading indicator along with the NS decay factor, and a LASSO model that combines NS factors with macroeconomic indicators and oil market specific predictors. These models significantly reduce forecast errors relative to a no change benchmark across a range of return horizons and futures contract maturities. We also find consistent evidence that models that use the NS factors result in trading strategies with higher Sharpe ratios and better skewness properties than buy and hold strategies and historical mean strategies.