It is naive to use Gauss Markov theorem. First of all we are talking about prices of an asset. There is extensive literature showing that we are far away from Gaussian behavior.
Second, all what the power law model implies that there is a general trend that looks like a power law, in fact the general trend looks like a straight line when plotted in a log-log plot. It is pretty obvious and striking.
No other asset behaves like this and I tried many (metal, stocks, forex).
It is amazing we get an R² of 0.92. It is easier to get high R² with log data that is true but it is high remarkable.
We need to do more sophisticated analysis to determine how close to a real power law we are.
It is also obvious that there are large deviation from the trend. The bubbles are such deviations by definition.
So we need to study the Bull runs and Drawdown separately.
What is their behavior? Do the occur regularly? What is the price distribution during these times? How the differn from underlying trend?
These are the interesting questions and not meaningless Gaussian tests to indicate the goodness of fitting.