Chris;

That's not how it works.

I evaluate predictions and one quake is enough for a hit. I don't give credit for multiple quakes unless the number is predicted.

The odds on a hit works the same way. I divide an appropriate time period into consecutive windows of the predicted size, drop the proper quakes into the windows and count how many windows have one or more quakes. That divided by the number of windows is the odds on such a window getting a quake.

The point in doing it this way is to eliminate the effect of clustering.

Roger