It seemed ten or fifteen years ago, when we tracked a storm for about a week, and it then showed marked improvement, we were pretty confident that we would score with that storm with a lead time of 5 to 6 days. Now the models seem to be all over the place up to a couple of days before game time. Frustrating.