Although we do not know their method, I was able to find a method that is darn good at getting the same numbers. I took the top 20 non-season worldwide clears, subtracted 10 GRs, and had a fudge factor to reflect the difference in DPS attributed to the average paragon levels of the top 20 clears. The fudge factor was ~0.8 GR level difference between highest and lowest paragon classes.
Given this, I think the top clears are still relevant. The problem with the PTR is projecting true build potential. The current PTR wizard GR 150s is due to a bug. Even if GR 150 was clearable without the bug on the PTR, Blizzard would definitely nerf the build. A rule of thumb is that the top clears on live always exceed new builds on the PTR (assuming no changes between the two) by 4 and often more GRs.
P.S. Whatever their method is, it has a serious flaw. There is no way that a 5K witch doctor with the pandemonium buff would only clear GR 120 while in non-season would clear GR 130 as their tables reported in the game balance blog post.