CMStatistics 2022: Start Registration
View Submission - CFE
Title: Power enhancement in detecting sparse signals, with applications to correlated test statistics in finance Authors:  Nabil Bouamara - UCLouvain (Belgium) [presenting]
Sebastien Laurent - AMU (France)
Shuping Shi - Macquarie University (Australia)
Abstract: A simple tool is introduced to control for false discoveries and identify individual signals when there are many tests, the test statistics are correlated and the signals are potentially sparse. In such situations, the Cauchy combination test aims for a global statement over a set of null hypotheses by transforming and summing individual p-values. We unravel the combination test to find out which of the p-values trigger the global test to, for example in the context of time series data, timestamp rejections. We also revisit two multiple-hypothesis testing problems in financial econometrics for which the test statistics have either serial dependence or cross-sectional dependence. We conclude that using the raw p-values in another way boosts the power compared to the workhorse procedures and bypasses a lot of the drawbacks inherent in extreme value tests, quadratic forms, thresholding and screening techniques.