[GGG]

Originally Posted by #-3

I don't mind the methodology of trying to say in general team X did better then they should of, and team Y did worse than they should have. As a predictor it does not take into account player movement, development, or regression.

Trying to put his methodology to the test, I found.

the 15 luckiest teams 08/09, had a net change of -55 point in 09/10. 10 did worse
the 15 unluckiest teams 08/09, had a net change of 74 point in 09/10. 9 did better

the 15 luckiest teams 09/10, had a net change of -43 point in 10/11. 9 did worse
the 15 unluckiest teams 09/10, had a net change of 39 point in 10/11. 10 did better

the 15 luckiest teams 10/11, had a net change of -60 point in 11/12. 11 did worse
the 15 unluckiest teams 10/11, had a net change of 63 point in 11/12. 8 did better

the 15 luckiest teams 11/12, had a net change of -75 point in 12/13. 9 did worse
the 15 unluckiest teams 11/12, had a net change of 136 point in 12/13. 9 did better
* pro-rated the points to an 82 game season.

So historically about 66% of time he successfully predicts if a team will improve or regress. Not bad, but I think most of us could go about 60% just on gut feeling.

Which stat are you using for luck. If it is straight PDO then this fits what the article was saying where he was taking 3 parts league average shooting and 1 part team shooting and 2 parts league SV% and 1 part team SV%.