One simple technique would be to find a distribution that is fairly independent of trading strategy and then do DistributionFitTest[expectedDist, observedDist]
The expected distribution might be an EmpiricalDistribution established from past data, or something theoretical like BenfordsDistribution on the trade size.
The toy example I have used for school kids is to detect fake coin tossing by looking at run lengths (should be exponentially distributed) and partitions of the data into 2s or 3s (should be uniformly distributed).