Michael;

much better; I can read it now.

Where did this site go? I haven't been able to get on for the last several days.

I've reviewed your earlier post on probabilities and there are some problems.

You suggest overlapping windows. That's no good. It makes it possible for a quake to be in more than one window. What is required is consecutive windows. We can handle odd sizes by increasing the window size calculations and input but the windows must not overlap.

Neither can the predictions overlap.

So for a 70.5 hour window, we convert that to 2.9375 days and the windows are;

day 1 to day 2.9375
day 2.9375 to day 5.875
day 5.875 to day 8.81

and so on.

We still need probabilities for all predictions, hit or miss. It occurs to me that the score for a prediction is 1 - probability, so an unlikely hit gets a high score and an likely hit gets a low score. A miss gets a negative score and the overall average is used to determine credibility.

This is going to get into a lot of work. Are you up to it?

Roger