A Comparative Study of Multi-step-ahead Prediction for Crude Oil Price with Support Vector Regression

Abstract

Accurate prediction on crude oil price in a long time horizon has been appealing both for academia and practitioners. Recursive strategy and direct strategy are two mainstream modeling schemas widely used for multi-step-ahead prediction in the context of time series modeling. In this paper, a comparative study has been conducted to justify these two strategies in multi-step-ahead prediction for crude oil price with Support Vector Regression (SVR). The experimental results show the direct strategy has more consistent performance than recursive one in the various experimental setting.