The point is that Hogwarts houses are literally completely arbitrary, and not even an attempt for rigorously categorizing people who aren't fictional characters (and even many of the fictional characters that Rowling actually wrote could have plausibly been in a different house, some explicitly so, and many Gryffindors would have arguably been more appropriate in some other house).
As evidenced by the test I just took, I get a relatively high score in Gryffindor and substantially higher than zero scores in Ravenclaw and Hufflepuff, and they all add up to a lot more points in total than what I have in Slytherin, so if you compress that data into a 2-bit value which is what the house system does, all of that information disappears even though it's the majority of the information that was available before the compression. And I would expect that my Slytherin result is actually a lot higher than what the average person gets in any house, so for the vast majority of people, at least as much information gets lost in the compression. The average person probably moves houses if they keep retaking the test every few months.
This is all transparently ridiculous. If you have to admit that the test nonetheless meets all the same standards that you use to evaluate the MBTI highly, that should probably be a very strong sign that those standards are not high enough.