Multi-perspective investor attention and oil futures volatility forecasting

Recent literature discloses that crude oil price co-moves with alternative energy prices, has volatility spillover effects with related markets, and is affected by macro-economy and geopolitics factors. Thus, we propose to construct multi-perspective investor attention index based on Google Trends considering these four perspectives besides the common perspective of crude oil, so as to augment oil price volatility prediction. Our empirical results show that, investor attention to different perspectives contributes differently to oil price volatility over different forecast horizons, and the model incorporating multi-perspective investor attention significantly outperforms the benchmark volatility forecasting model and the model incorporating only investor attention to crude oil. Moreover, considering heterogeneity and asymmetry of multi-perspective investor attention further contributes to oil price volatility prediction, and combing various attention-augmented models with the dynamic model selection approach leads to forecasts with stable and significant superiority.