Not directly addressing this, but perhaps the point is not to create a model to address every known limitations. No models are perfect. Points are made based on a model, and a limitation/assumption is pointed out. You need to judge given the assumption/limitation if the message made still some-what make sense.
I mean something X being highest right now is a pretty clear statement, it is either true or false, if X is well defined. But then your problem is you want to have Z instead, which hopefully is accounted for every other effect except for the factor you want to measure, such as inflation, GDP, etc. But the more you add, you either makes your Z definition sensitive to these other factors you’re not interested in (such as a peak comes from the denominator, not the numerator), or we really admit we can’t calculate Z for sure, and is now a statistical model (any kind of smoothing including the MA you mentioned is to admit we don’t know the true state but can only observe it with noise.) Now the statement of “Z is highest now” is no longer a true or false statement, but a probabilistic one.
Certainly then P(Z is highest) = 60% or so wouldn’t make much of a headline and can be confusing.
So my point was rather than trying to create a perfect model (which you can’t, and what you learn from it won’t be clear as well), sometimes just deal with the statement and methodology and judge if it is a clickbait or has some merit of truth there.