Sort:  

Historical data might fit nicely a lot of the time but it's what happens after the convenient fit that's important.