I agree with Templars. His seeding is standard and has several nice features:

1) the "average position" in each group is #16.5

2) seed #1's best starting opponent is #8,
#2's is #7,
#3's is #6, and
#4's is #5.
This gives a slight advantage to #1, as it should be.

3) seed #1 can meet seeds #2 and #3 only in the final (I think this does not apply in our case, though, unless you are planning to let 4 players move up from each group).

This said, I guess the stats you are using are flawed anyway, because in the last months only kills were awarded, and no deaths... so screw you and your tournament