Scratching my head here...

What is wrong with 3-5 day windows with well defined dates in general?

What is wrong with a minimal magnitude?

What is wrong with putting in measures designed to prevent gross artifical inflation of one's score?

Basically, the argument that has been present in prediction and evaluation for quite some time is accountablity...everyone wants it - but when it comes time to nail folks down with goals, objectives, criteria, and operating parameters suddenly it's as if everyone "whining" starts backpedalling with, "too restrictive"...or "no one would want to participate" or some other kind of "nuance" where people say what they don't mean or mean what they don't say...

Just thought of a brilliant idea...

Let's allow only 1-day windows.
Let's allow a radius of only 100 meters.
Let's require that a predictor predict the magnitude to the hundredth.
Ban anyone who says, "I was close!"

All in favor say "I"...

Otherwise, please mean what you say, or say what you mean. I do try to take people seriously...I really thought my suggestions were reasonable and allowed plenty of wiggle room already...

Note for Roger:

Sounds like you are trying to do what Mapquest/Yahoo maps service does...

Might want to do a google search on how they do it alogrithms wise.

=8-)