Highly accurate interval forecasting of a stock price index is fundamental tosuccessfully making a profit when making investment decisions, by providing arange of values rather than a point estimate. In this study, we investigate thepossibility of forecasting an interval-valued stock price index series overshort and long horizons using multi-output support vector regression (MSVR).Furthermore, this study proposes a firefly algorithm (FA)-based approach, builton the established MSVR, for determining the parameters of MSVR (abbreviated asFA-MSVR). Three globally traded broad market indices are used to compare theperformance of the proposed FA-MSVR method with selected counterparts. Thequantitative and comprehensive assessments are performed on the basis ofstatistical criteria, economic criteria, and computational cost. In terms ofstatistical criteria, we compare the out-of-sample forecasting usinggoodness-of-forecast measures and testing approaches. In terms of economiccriteria, we assess the relative forecast performance with a simple tradingstrategy. The results obtained in this study indicate that the proposed FA-MSVRmethod is a promising alternative for forecasting interval-valued financialtime series.
展开▼
