Testing Investment Strategies for Superior Predictive Ability

When different models are tested on one data sample and repeatedly altered in order to be found significant, the results are likely spurious. This is data-snooping – an ever-growing problem in the finance industry likely due to fierce competition and developments in data processing capacity. In academia, although recognized as a deplorable practice, data-snooping is likewise pervasive perhaps as a result of poor incentive structures at both the university and publisher levels. I manifest the problem of data-snooping through multiple academic and industry examples and then summarize Halbert White and Peter Hansen’s offered solutions, White’s Reality Check and Hansen’s Test for Superior Predictive Ability. I demonstrate the application of their tests by examining several passive investment strategies applicable to recent market moves and report my results