These stats should not auto-ban, but flag for human review.

ImO, the goal is to find the statistic cheaters will skew since they are cheating.

Lets say, one is wall-hacking in Counter-Strike. In Counter-Strike, many maps have a central location/lane called "mid". There is often a kind of mind-game of limited information going on:

  - Does 0, 1, or both teams have an AWP(=one-shot kill sniper)
  - Is the enemy sniper guarding mid?
  - Which entrance to mid is the enemy sniper currently watching?
  - Has the enemy sniper guarded mid in the beginning, but has now left?

Any wall-hacker will be a major outlier regarding not running into mid while the enemy sniper is watching.

If you can accurately dissect/classify/log player behavior, I expect the statistics to look similar to those of poker players who cheat and know the cards of the other players on the table: https://media.pokertube.com/uploads/2025/12/5d96fe17d39043a0...