Methods currently implemented for the calculation of potential powers are not state-of-the-art.
It would be great if more methods were implemented, like the power-to-power method that Greenbyte wrote a white paper about a couple of years ago, methods involving Machine Learning, methods using learnt power curves computed based on wind speeds from a meteorological model, learnt power curves by wind direction sector with biased nacelle wind speeds measured during stops being rebuilt from data of reference turbines, etc.