You fail to grasp a fundamental reality: data about the past can only tell us about the past. Data about the present can be useful for making predictions. Some of it is so useful that using it is criminally prosecuted by the SEC.
By "naive" I mean any approach that relies solely on trends rather than an actual understanding of underlying causal factors. Sadly there are so many of those factors that a correct analysis is virtually impossible, but that's no excuse not to try since some factors tend to dominate others and those can sometimes be found.