"If you want to make price predictions, don’t use regressions, but just calculate some growth rate and extrapolate it into the future. Please consult my previous article on this particular topic." That is what physicists do when they try to extract the power of the power relationship. The slope will give the power n so you do the linear regression for that. The power of a power law tells you a lot about the nature of the relationship and you can derive all sorts of important properties for the system. Also it is telling us the rough trajectory that it can be extrapolated. I did that myself in 2018 (the article Burger references and took the idea from) and today the behavior is basically the same with similar parameters. The model also give us regions where the price is overbought or oversold relative to the general trend and the deviations of about 40 % from the trajectory allign almost perfectly with the bottoms and bear market. It is a very useful model, it doesn't have to be right at the statistical rigor level you are trying to test it.