#2 wont work for bracket one of each server cluster, there aren't enough 80k+ to have a sage with at most 5-7k difference.
At the same time, for latter servers, the servers that die tend to be due to super ptw making the game unplayable in sage like you said, so quite a few of those are otherwise stranded and need Xserver, therefore....

This is only partially true, because some of the stat cancels out with each other, like atk vs def, nin vs res, crit vs crit etc.
5000 crit vs 3000 crit, for example, might be exactly the same as 2000 crit vs 0 crit (I'm sure I don't know how the actual formula works)
So when all is said and done, 100k vs 80k might be closer to 50k vs 30k than you might think, except both side gets a bigger hp buffer so it won't be a quick crashing battle... maybe..

So chances are, the effect of stat difference is probably no linear to the gap nor the % gap. I would guess something in between but I'm sure I don't know.

Since the Xserver stuff is based on same cluster as space time, you can easily check how many active player of what power range there is.

For example, S2-S12 LA have exactly 11 80k+, certainly not enough for a sage. Top player is 110k, so that's quite a bit of a gap. To get say 39 players for a decent sized sage, it would go down to roughly 67k or 68k. And that's assuming all of them sign up, realistically probably 65 or 66k will get drafted if it's top 39.
And given there are 11 80k+, which means quite likely 9 of them show up in sage, they are the most likely to win rewards. And even if they don't chances are some of the dozen or so 70k got them and the high 60s are getting nothing.

Hard bracketing will simply never really work out for group 1 of each server cluster...


On a random note, here is a system that could potentially be "fairish"--make group 1 "top 9 power and all top 3 winners of last sage(9 per field)". Pretty much means that no one is gonna win sage twice in a row except the top 9 power (which are likely to win anyway so whatever).

Amd with the top 9(possibly some of top 15, since top power aren't guaranteed to win so some of the others might) "removed" the power gap we are left to work with is much less.

For example, for my server cluster, the next 39 or so will probably be around 64k to 75k or so(again, some ppl wont sign up), which is still large but much more reasonable. And the gap would only get smaller in the following groups.