Q–Q plots

Is a metric’s distribution normal? A normal Q–Q plot compares each metric’s empirical quantiles (exact, from seedDistributions.bin) against the quantiles of a normal with the same mean and standard deviation. Points on the diagonal mean the metric is Gaussian; systematic curvature reveals skew or heavy tails.

Most Sage metrics are sums of many independent PRNG draws, so by the central limit theorem they should sit close to the line — any deviation is the interesting signal.