IMO, close misses should count as hits. Say somebody had a prediction out for an M7 to hit within 10km of LA in the next 3 days. Instead, LA gets hit with an M6.8 at 20km from LA at day #2. Does that prediction deserve a miss? I don't think so.

Instead I think that we need to go to a point system where each of the three (3) parameters of location, size, and time frame are each assigned a point value. The three point values would be added together and then averaged where an average of 85 or above would be considered a hit. So for each of the parameters we could go with:

time frame:
100 points - within window
90 points - within 1 day
80 points - within 3 days

magnitude
100 points - as predicted
90 points - within .2
80 points - within .5

location
100 points - within 25km
90 points - within 50km
80 points - within 100km

So for the above scenario, it would have had 100 for the time frame, 90 for the magnitude, and 90 for the location for a total of 280 averaged at 93.3 a hit.

In my personal opinion, without doing something like the above, prediction will never get anywhere.

Dennis