Pattern and determinants of tail-risk transmission between cryptocurrency markets: new evidence from recent crisis episodes

Financial Innovation

Table 9 Out-of-sample prediction (MSFE-adj.)

The table presents the Clark and West (2007) mean square forecast error (MSFE)-adjusted statistic comparing the out-of-sample predictions. The in-sample period is taken between 11/10/2017 and 5/30/2021, while the rest is considered as an out-of-sample evaluation forecast period. The results are reported for the forecast horizons ∈ {10, 20, 30}. The null hypothesis of the Clark and West (2007) test is that the model has no predictability; the model containing information on investor sentiment/economic condition does not improve the predictions. Values reported in square brackets are the t-statistics. According to Clark and West (2007), if the t-statistic is larger than + 1.282 (for a one-sided 0.10 test) or + 1.645 (for a one-sided 0.05 test), then the null hypothesis should be rejected. The asymptotic critical values; *** denotes statistic significant at the 1% level (see Clark & West, 2007)