In this paper, we develop a new approach to test whether momentum is indeed an anomaly in that it reflects delayed reactions, or continued overreactions, to firm specific news. Our methodology does not depend on a specific model of expected returns and, more importantly, does not require a...