My usual approach is to try to get close to the desired look and do the rest with the tools of NLP and/or Lightroom Classic. I never try to get colours “correct” in a scientific sense, but to make them look like I either remember or want.
Kodachrome (and to a lesser extent) Ektachrome were not formulated to deliver true-to-life colours and tonality, but to have a look that might please customers (saturated colours?) or meet a situation (night sky?) or requirement (fine grain) etc. There was nothing “objective” in these slides.
A preset created by optimising a single slide or several slides of one film might help to get close to target, but then again, it might not. NLP is made to adapt to the images it converts, which means that a change of tonality or colour cast or a different balance of colours and tones can make converted images look different. Old slides (and negatives) tent to emphasise such differences.
Try to select the settings that suit you best or create a self-made preset, bearing in mind that such presets work, but not in all cases. My preferred way to explore a technically difficult photo (negative or positive) is to use virtual copies to which I apply e.g different colour models vs. different pre-saturation values. One example of such a systematic approach can be seen here.