As I see it, scientists have three categories of non-science.
Theories that actually have been falsified with data showing they are wrong.
Theories that cannot be tested regardless of technological advances, e.g., anything that by definition is non-observable like another universe isolated from ours. This is philosophy in fact.
Theories that cannot be tested because statistics can never draw a conclusion about it.
The difference between 2 and 3 is that 2 has no predictive power. As long as it has, it can be theoretically tested, maybe in unreal conditions. In 3 the predictive power is very small, but not zero. Therefore I'm not sure we can really say it is unscientific.
However, regarding the last position I often hear that since you cannot use statistics to show the theory to be true, your theory is not significant or even meaningless. Nevertheless, suppose I have a unique experience that no one else ever had. Why can't I share this experience? I cannot use telepathy. I don't know if others have the same experience without sharing it. But this experience can affect me. I can make predictions about it, but they are too localized.
But what if after I telepathically shared my experience people agreed my theory is not meaningless? So, is it really correct to say the theory is pseudo-scientific if its predictive power is too small?