Good point. It's kinda hard to determine which term length is the right one though. Should all history necessarily be considered? If not, what part of it is relevant? I'm not completely sure.
The answer is none of it.
You can't predict future price that way.
Well, maybe YOU can't.
TA is still fun and there is tons of empirical evidence that its fruits, when properly backtested, can provide steady profits.
Even if said evidence did not exist, people would still do it if they acquire enough enjoyment from the activity.
No matter how much you tell people "you can't" about ANYTHING, there will always be at least someone who will attempt to do that thing.
Indeed. Those who doubt TA's applicability often stand behind the myth that one cannot "predict the future". Yet, we do it all the time. I can tell you exactly what day this solar eclipse will be on, for instance. Some things are harder to predict than others, but like everything that isn't quantum they are still predictable.