A miss for a large earthquake as far as time goes should be allowed a slightly larger margin of error than that of a small mag event. I suggest a two tiered "prediction miss rating system", where you classify the prediction either major or minor. Minor forecasts have much stricter scoring, three days off, max. Major quakes that may occur on scale of years apart should be given more leeway in scoring. A month off on a mag 7.4 that only occurs each 50 years is remarkable. A month off on a 3+ in Hollister means you're a couple of quakes behind.
Place scores- Minor quakes should be 50 miles or depend on population density. More towns nearby means less leeway as to range from. Major quakes can have more range allowance, and a larger margin for error to still score something.
Magnitude-should never be more than two degrees as the difference between two degrees is 1000 times energy released. With smaller quakes, one to one point five would be about right.
A prediction format would serve it's users best if it analized what went wrong and by how much, for this will teach us to improve our weak spots. A board that merely says Hit or Miss is more an exercise in frustration especially when we are going to get more close calls than hits at first. If one can see they made a close call they are more likely to post again and not quit frustrated. I already have two hits on the board that aren't showing up yet and 6 misses/near misses that are showing up.(I know you are busy) I had a hit up for a few seconds before it got archived as the last entry. Bummer. I worked hard on that and it's gone to never viewer land...mb.