there always is one and only one best-fitting trendline for USD/BTC all time dataset
Mathematically false. For example, there may be only one least squares linear fit (on a logarithmic scale), but that is just one model and you do not know which model is the correct one. For example, I posit that log-logistic curve fit is more apropos.
A trendline with lesser fit can have greater predicting power. But there are no more than 1 trendline with the best fit, and it is mine. Situation is analogous to our shoe size, either I have larger, or you have, or they may be the same, but only if they actually are the same.
How can you assert that a fit with one model is lesser fit than a fit with another model? Define 'lesser'?