Brian;

They give the same answer but it looks different.

The result my program gives is 0.546 for 226 predictions and 71 hits using a 3 day window on 6.5 mag predictions.

The odds are 0.300 for such a hit and z-binomial gives 0.348 which looks to be different.

The reason is that the z-binomial test gives area of the normal curve inside or outside the point while my program gives standard deviations from the midpoint of the standard curve.

When I use Amit's real mag predictions he gets fewer hits and falls slightly below chance.

Roger