[ QUOTE ]
now you retards woke me up, and I'm gonna have to cut through this BS with some real math. If you have two sets of player data at those two limits, measured by those stats you mentioned, you are basically modeling players as some n-variate random variable from two unknown distributions. Your data sets are basically a set of samples from those two distributions. Now, if you had some way of estimating those two distributions, one standard of measuring the "difference " between those two distributions is something known as "Kullback-Leibler divergence" (justfukingoogleit), and if you want to use the empirical distributions as estimates of the real distributions, you could take the KL divergence between the two of those to measure the difference in play at those limits. Otherwise, you guys can stuff your step functions and interpolated splines back in your holes.
[/ QUOTE ]
