Such a study is practically inevitable. Studies generally need central tendency to be statistically significant. Long trends are kind of by necessity outliers.

Such a study is practically inevitable. Studies generally need central tendency to be statistically significant. Long trends are kind of by necessity outliers. Both are valid and depend on your modeling and time horizon. But if you can find a simple statistical alpha which has more examples, you're probably better off doing that. Which is why HFT was so profitable.