Oil Price Forecasting with Hierarchical Multiple Kernel Machines

The dynamics of oil prices are nonlinear and non-stationary. They are also tightly correlated with global financial markets. Traditional models are not very effective in forecasting oil prices. To address the problem, this study employs a new kernel methods-hierarchical multiple kernel machine (HMKM) to solve the problem. Using information from oil, gold, and currency markets. HMKM exploits multiple information sources with strong capability to identify the relevant ones and their apposite kernel representation. Empirical results demonstrate that our new system robustly outperforms traditional neural networks and regression models. The new system significantly reduces the forecasting errors.