U20 tournament - Junior Evaluation Camp (Sieloff, Gillies, Monahan, Poirier & Janko)

[edslunch]

Originally Posted by ThePrince

Yup, agree with all of this. What you don't realize is that with enough data points in the set, these effects should be averaged out. If there are 10 players playing in an offensive system increasing their scoring, and 10 players playing in a defensive system decreasing their scoring, these effects should (very roughly) cancel out. With a lot of data points, you get even more accurate results.

This is exactly the issue I have with The use of NHL-E. It 'deals' with differences in philosophy, age, role, etc. by averaging everything out. As you get more data points you get a better average. This makes it a useful rule of thumb for comparing leagues but means that virtually every player is an outlier in one way or other. So putting too much stock into it for analyzing.

[Tinordi]

Originally Posted by edslunch

This is exactly the issue I have with The use of NHL-E. It 'deals' with differences in philosophy, age, role, etc. by averaging everything out. As you get more data points you get a better average. This makes it a useful rule of thumb for comparing leagues but means that virtually every player is an outlier in one way or other. So putting too much stock into it for analyzing.

Meh this is a mostly academic critique. Until we understand the variance of the data we can't criticize NHL-e for this problem. If we know the mean point equivalency per league, we could look at what two standard deviations are in points and make better conclusions.

I don't know what the standard deviations are between the leagues for NHL-e. Anyone have an idea?

[BACKCHECK!!!]

Originally Posted by Tinordi

I don't know what the standard deviations are between the leagues for NHL-e. Anyone have an idea?

There is no gold-standard for measuring NHL-E. It depends on each league, how many players are moving between the leagues, and how large of a time sample you want to use.

So every time you calculate an NHL-E for a given league over a given time frame you need to calculate a new SD. I've seen some people include them in their results, some people don't.

For some leagues (ie the AHL and CHL) they're pretty tight (just guessing I'd say they're probably around 0.1 PPG or even less).

[Fire in the disco]

So... Any news abouts camp??

[Tinordi]

Yeah so if the standard deviations are tight then eds criticism doesn't really stand provided there's enough observations/degrees of freedom.

[Vulcan]

For the AHL, the NHL- E is skewed because teams aren't calling up the best scorers but are calling up players to fit a certain role. This lowers the AHL score as we often see 4th liners and pugilists being the first call ups.

[Coach]

I keep coming in here for the slight hope that there might be....prospect news...

[Alberta_Beef]

Originally Posted by SebC

Only if you can show us a correlation between penis size and being right on the internet.

I bet it would have something to do with growth