Nonparametric monitoring of equal predictive ability |
| |
Authors: | Vasyl Golosnoy Yarema Okhrin |
| |
Institution: | a Institute of Statistics and Econometrics, University of Kiel, Olshausenstr. 40, 24118 Kiel, Germany b Department of Statistics, University of Augsburg, Universitätsstr. 16, 86159 Augsburg, Germany |
| |
Abstract: | This paper introduces a novel sequential approach for online surveillance of the equal predictive ability (EPA) hypothesis presumed to hold for many competing forecasting models. A nonparametric control chart is suggested for providing a decision at every new time point as to whether the EPA hypothesis remains valid. The detection ability of our procedure is evaluated in a Monte Carlo simulation study for various types of deviations from the EPA hypothesis. Our approach enables the quick detection of various shift types, is parsimonious, and robust to misspecifications. Based on these results, we formulate practical recommendations for procedure design. |
| |
Keywords: | EWMA control charts Nonparametric control charts Ordinary sign test Wilcoxon signed rank test |
本文献已被 ScienceDirect 等数据库收录! |
|