I've done the same you have for longer, I've worked through research paper after research paper. I've gone through system after system. I type people inside multiple systems pretty high - as in, objectively fairly high because I can confirm it with formally tested people. I've derived my own shortcuts, just as you have. I've gamed systems that I haven't even explored before. Personality is something I know inside and out. The point isn't that I can do it better than you can - the point is that neither is a "system". I'm not right because I have done this... Instead, I take the divisions within MBTI (since we are using MBTI - I rarely use it) and match up behaviour against a set model. A system must be validated to know what to look for. My own personal biases get in the way all the time - I don't have enough people to type to know if I am correct, I don't know if behaviour is conditional, I can't observe enough in order to be reliable. I am very aware of what biases does inside type (post #29 in my sig, if you are curious). What you need is to validate how you determine what someone is - at least, if you want to be credible. Doesn't matter for personal use.
That's what MBTI does and that is what the research is based upon. You discredit MBTI for not being personal, ignoring that you are one subjective person observing behaviour... while MBTI is a test with a solid million examples and interviews. You attack the verification of the test, the research and methodology, research that I very much doubt you have even read.
Let me be clear - I don't even support MBTI. I am a major opponent to MBTI. But I don't see how you can cast stones at MBTI when you offer no alternative - except your personal opinion on what people are. MBTI is simply the validated version of your opinion... and of course, it is closer to the source - it actually asks preferences rather than guesstimates from behaviour.