I'm not sure if gravedigger is still interested in this topic, but i would like to chime in.
There is no 100% way to tell a good cheater apart from a non cheater solely by stats, with that said there is a margin for error because skill can be interpreted many different ways. Lets say you take the #1 player in the world and post him in a 1v1 with a notorious hacker that knows his hacks inside and out but also has a higher than average skill in the game. This setup should result in the hacker winning just about every round. Due to his natural skill in the game being above average and the assistance of the hacks. Now lets put that same #1 player against someone that does not have the skill level nor the knowledge of the hack, this should result in the #1 player winning just about every round. HOWEVER! if the hacker is not trying to hide his self then at this point the tables will turn and no one will beat him.
On the flip side of this stats that are 200-300% above average of a database as large as BF4DB, FF, GCC, PBBANS, ...ect have to raise an eyebrow or two, due to the fact that at 300% that is 3x higher than the baseline of 200k-300k players ... this is why there is a tiered system in place giving a cushion between the tiers Green being avg - +100% Yellow being +100% - +200% and red being 200% - 300% (these numbers are just an example and may not be 100% correct to BF4DB)
