Quote:
Originally Posted by Jay Random
Assuming the inputs to be average, when most of the inputs are AHL-level goaltending prospects, is not better than no analysis at all. It's worse.
|
With sufficient sample size, inputs regress to the mean. We don't need to look at inputs to know that Edmonton is no good. A goalie coach whose goalies put up good numbers is likely to be better than one whose goalies do the opposite, even if we don't know who their goalies are; it's just not as likely as a goalie coach who improves his goalies being better than one who makes them worse.