Fig. 6

RMSE loss function for \({Y}_{q+1\backslash q}^{\left(0\right)}\) and \({Y}_{q+1\backslash q}^{*\left(0\right)}\) along with the dispersion measure, \(DM.\) The RMSE for the one-step-ahead forecasts of \({Y}_{q}^{\left(0\right)}\) from the estimated regression model: \({Y}_{q+1\backslash q}^{*\left(0\right)}\) is \(RMSE=\sqrt{{10.000}^{-1}\sum_{q=1}^{10.000}{\left({Y}_{q+1}^{\left(0\right)}-{Y}_{q+1\backslash q}^{*\left(0\right)}\right)}^{2}}\). The RMSE for the one-step-ahead forecasts of \({Y}_{q}^{\left(0\right)}\) from the aggregation of the forecasts: \({Y}_{q+1\backslash q}^{\left(0\right)}\) is \(RMSE=\sqrt{{10.000}^{-1}\sum_{q=1}^{10.000}{\left({Y}_{q+1}^{\left(0\right)}-{\sum }_{k=1}^{4}{Y}_{q+1\backslash q}^{\left(k\right)}\right)}^{2}}\). The \(DM={\sum }_{k=1}^{4}{\left({\beta }_{1}^{\left(k\right)}-\overline{{\beta }_{1}}\right)}^{2}\), for \(\overline{{\beta }_{1}}={\sum }_{k=1}^{4}{\beta }_{1}^{\left(k\right)}/4\). The DM is presented on the right-hand-side axis