Resulting translient statistic (Z ²; left panel) and proper statistic (S ²; right panel) images computed for zero-mean noise-only images (no moving point source). These patterns (arising from the action of the translient statistic and proper statistic on the observational noise) are responsible for false-positive detection events (at a certain detection threshold), i.e., cases of an event being classified as a detection of a translating point source when there was none. The highest peak, on the left panel, (Z ² ≅ 11) corresponds to a false-alarm probability of 0.004 pixel^–1 (i.e., a χ ² distribution with 2 degrees of freedom) and close to unity for 64² trials.