Define reliable. If a metric had 95% accuracy would that qualify as reliable? I would certainly say so. Especially if the 5% outliers are not random but in fac

Define reliable. If a metric had 95% accuracy would that qualify as reliable? I would certainly say so. Especially if the 5% outliers are not random but in fact an extremely obvious edge case? I do not know how “reliable” BMI is. But elite athletes represent an exceedingly small percent of the population. My intuition is that BMI is a pretty darn good first approximation. And instances where it’s not good there is supporting evidence.