Quote:
Originally Posted by Ol' Jase
???
It's 82 games, and exactly the same sample size as all of the other stat sample sizes.

There is more to sample size then the number of games. Its about trackable events. The oilers had 306 goals(either for against) at 5v5 last year that go into GF%(which is really just goal differential expressed differently).
Compare that to 5224 Corsi events for the Oilers last year, and yes Goals is far to small of a sample size.
To put it another way, its pretty much common sense that a team with a good GF% is going to make the playoffs. Teams with a positive goal differential rarely miss the playoffs.
The question guys like Dellow and Eric Tulsky are asking and trying to determine is why those teams have good goal differentials, and what aspects of that good goal differential are repeatable.