Hi Roger:

I'm confused. Last post you said it was more accurate, this one you said it wasn't. Which is it and why?

Who makes odd sized predictions? Just take a look at the Prediction Status Table. Looks like about half of the prediction are odd size, so I would have to say lots of predictions are odd sized. Clearly it's something that has to be taken into account. I also don't think predictors should be forced into a set time frame created by us. It's their gig and they should have as much free reign as possible. Setting limits will only serve to stiffle experimentation.

Granted all methods will makes some assumptions, but if there is a way to create more accuracy, it should be done, don't you think? With the example I provided you with the 100 day catalog, depending how I looked at it I could cause a 100% "error" rate, which is a bit high for me. Granted, 1% or 2% is not much a difference when looking at the big picture, but when somebody asks me how I came to that number, I would like to get it right the first time so I don't have to have this same discussion with the next "me" that comes along ;)

Michael