The beauty of the 5.25-year trendline with exponential fit (R^2=0.93 which is pretty darn good) is that it takes into account every worry of every person who has ever owned or not owned bitcoins. I put more weight on that than the individual worries of a single person.
The trendline is a representation of how people have speculated in the history of Bitcoin. As Anonymint points out we have a large merchant adoption that we haven't had the last 5.25 years. Do you think new factors with a possible large impact like this can make the trendline less accurate?
The more transaction traffic, the more accurate the models should become. The trendline can be derived from the active wallet trend. As sample size increases, moments computed on the sample convert to the moments of the population. It's only when the samples start coming from a different population that the trendline would be less accurate. That would happen if market dynamics underwent a major structural change.