Thread: Vs#2 Thread
View Single Post
02-11-2013, 12:50 PM
seventieslord's Avatar
Join Date: Mar 2006
Location: Regina, SK
Country: Canada
Posts: 26,525
vCash: 500
Originally Posted by Bring Back Scuderi View Post
Let's talk Defensemen scores in the early 70s for no other reason than I'm doing them for Stapleton.

Bobby Orr in 1970 is my first outlier (surprise) and that carries through 1975. I'm using the first non-outlier as my #2, but I don't know what to do with Brad Park in 1972 and 1974 (doesn't affect Stapleton, but should be discussed). It looks like a clear outlier to me, but I'd rather bring it to the group.

Here's the placements:

Orr - 117
Park - 73
Tremblay - 57
Guevremont - 53
Lapointe - 49

Orr - 122
Park - 82
Redmond - 59
Vadnais - 59
Potvin - 54
Smaller sample size, so much more room for multiple outliers. I like to look at where the "pack" begins.

Originally Posted by BraveCanadian View Post
You wouldn't be saying yikes at using Mullen (a non outlier who came across his points honestly) if Brown hadn't conveniently missed 12 games and 20 points, getting him to 135. Then it would be obvious that using him as the standard for comparison in a percentage system is a horrible idea. The fact that he "only" got 115 points makes it easier to dismiss his outlier status, regardless of the fact that he sat in the back of the bus.

The point of removing gretzky and lemieux is because no other era had players who dominated this much offensively season after season. If you keep them in, you hold the savards and hawerchuks to an impossible standard. So obviously you must remove them. The 1980s without those two, are fairly comparable to most other decades so it levels the playing field. But if we do that then their residual effects must also be accounted for.

That said, from my end, all this discussion is geared towards having results that "work" and can be used as a shorthand. I think what most of us would like to see is a system where the number of players reaching certain benchmarks gradually increases over time. I did some quick formulas using the results in my sheet and how each season looks compared to how it "should" look on the time.curve, and it appears that from 1980-1999 the most unforgiving seasons, in order are:

1999 - this was a crap year and results should lag a bit, but half as many 60 scores as 1984 makes little sense
1995 - might make sense if the tie for 1st is treated as one player making the standard 65 instead
1993 - uses oates as the standard (#2 non lemieux) but that is a bit harsh especially since this was the season of offense and if anything results should look extreme this year, not subdued.
1982, 1981 - Nilsson (12 points ahead of 4th) and Stastny ( in 3rd with 31% more than 9th) should not be the standard. Weird year for top end scorers. Savages should be the standard if we're being logical here

The easiest are:

1987: not much can be done here, even if we used 2nd place Kurri that's only 3 points more than my current standard
1989: there do appear to be "too many" hitting 60, 70, 80 compared to surrounding years, so that is a good reason to consider using a higher standard. But at the same time it is worth noting that this is a unique year. Percentage wise, 30th is closer to 10th than any other season I could find.
1984: very 1987-like. Using coffey instead would only help marginally.
1992: I should use Hull and his 109, not Messier and his 107 here. This year would look perfect.
1994. Very 1989- like with the punching at the 2nd tier. Ran numbers with fedorov as the standard instead of oates, but then the pendulum swings farther the other way. This year should be a bit heavier, lots of great players in their primes.

seventieslord is offline   Reply With Quote