IMO, statistical anomalies, however accurate, aren’t going to do much unless the mathematical probability is -0-, not close to -0- but -0-.
Isn’t this a civil case and settled by preponderance of the evidence?
Two or three standard deviations should surely be a preponderance.