There's a fascinating story about a probable tournament chess cheat. No one knows how he does it; there's only the facts that 1) historically he's not nearly as good as his recent record, and 2) his moves correlate almost perfectly with one of best computer chess programs. The general question is how valid statistical evidence is when there is no other corroborating evidence.
It reminds me of this story of a marathon runner who arguably has figured out how to cheat undetectably.