Bootstrap based multi-step ahead joint forecast densities for financial interval-valued time series

This study presents two interval-valued time series approaches to construct multivariate multi-step ahead joint forecast regions based on bootstrap algorithms. The first approach is fitting a dynamic bivariate system via VAR process for minimum and maximum of the interval while second applies mid-points half-ranges series. As novel perspective, we adopt techniques into proposed obtain lower/upper bounds intervals. forecasting performances are evaluated by extensive Monte Carlo simulations real-world examples: (i) monthly S &P 500 stock indices; (ii) USD / SEK exchange rates. Our results demonstrate that capable producing valid