If you're able to shift cheat detection systems to primarily focus on griefers, that's a significant win, since griefing by its nature can't be subtle enough that other people don't notice it. The whole point of griefing is to be noticeably toxic to other players and to get a reaction out of them. Griefers have a harder time hiding than other cheaters do.
If your design allows you to shift your focus towards trying to fight toxic behavior itself rather than just cheats, you can use the same detection methods as you would for any other toxic behavior that doesn't involve cheats. And you should have those detection methods anyway; whether or not a cheat is involved in the abuse shouldn't be the deciding factor between banning or allowing toxic behavior.
The shift here is in realizing that for a significant number of your players, playing against someone who legitimately snipes them from across the map over and over while camping at a spawn point because they're just that good, and playing against someone who uses an aimbot to accomplish the same behavior -- both of those experiences are equally game-ruining for a lot of players. The legitimacy isn't the problem for those players, the resulting experience is. So trying to detect subtle aimbots kind of misses what the real game-ruining problem is for a lot of players.
Where a number of player archetypes are concerned, you can look for overt toxic behavior and ban it regardless of how that behavior is accomplished. Similarly, if you're not prioritizing ladder integrity over everything else, there's a case to be made that you can just ignore any cheating that's too subtle for people to notice or that doesn't result in overt toxic behavior.
A nontrivial number of your players don't actually care about ladder integrity, they want a number to go up and they want rewards released on a variable schedule as they get better at the game. For those players, it's not clear that playing against cheaters is actually a problem unless the cheaters ruin the difficulty curve or ruin the matches, and... again, you should think about banning people who do that stuff even if they're not cheating.
If your design allows you to shift your focus towards trying to fight toxic behavior itself rather than just cheats, you can use the same detection methods as you would for any other toxic behavior that doesn't involve cheats. And you should have those detection methods anyway; whether or not a cheat is involved in the abuse shouldn't be the deciding factor between banning or allowing toxic behavior.
The shift here is in realizing that for a significant number of your players, playing against someone who legitimately snipes them from across the map over and over while camping at a spawn point because they're just that good, and playing against someone who uses an aimbot to accomplish the same behavior -- both of those experiences are equally game-ruining for a lot of players. The legitimacy isn't the problem for those players, the resulting experience is. So trying to detect subtle aimbots kind of misses what the real game-ruining problem is for a lot of players.
Where a number of player archetypes are concerned, you can look for overt toxic behavior and ban it regardless of how that behavior is accomplished. Similarly, if you're not prioritizing ladder integrity over everything else, there's a case to be made that you can just ignore any cheating that's too subtle for people to notice or that doesn't result in overt toxic behavior.
A nontrivial number of your players don't actually care about ladder integrity, they want a number to go up and they want rewards released on a variable schedule as they get better at the game. For those players, it's not clear that playing against cheaters is actually a problem unless the cheaters ruin the difficulty curve or ruin the matches, and... again, you should think about banning people who do that stuff even if they're not cheating.