Empirically, industry groups exhibit an exploitable momentum factor. How much of it is due to timing luck?