MindzEye
Wayward Ditch Pig
yeah there's been plenty of stuff written on why close is the best, which is usually under the umbrella of "score effects".
And it follows along with logic, too - teams adjust their strategy depending on the score. Teams will sit on leads, or even just plain let up with big ones, while teams a few goals down might just start riskng everything and going all out attack, or might just capitulate and give up. The Close numbers have for the most part been better predictors than the overall numbers.
the downside is the that you're cutting off a good chunk of your sample size when you do it and that those non-close situations probably do tell you something about the teams too.
So theyve started using score adjusted numbers that let you use all the data points without ignoring score effects. Of course, I'm not sure that they're doing quite the RIGHT adjustments yet, but the early returns seem to be sensible. They seem to make better predictions so far.....from what I know about them, which isn't a whole lot. its hard to find much info on the score adjusted numbers as of yet.
:cheers2: