Lowell;

That's why we need a cumulative score of some kind, one that considers all predictions by a given person. That way, hits count for you, misses count against you.

My original USGS program compared a person's cumulative results to a random result using the Kolmogoroff Smirnoff test for goodness of fit.

That's a bit much for here but if we take the inverse of probability and keep a running average it should be a pretty good indicator.

Roger