Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
atrich
Edmonton Oilers |
|
|
Location: Ottawa, ON Joined: 06.30.2008
|
|
|
|
|
The thing is... you didn't mention which teams. Look at the rise fall of lots of teams over the years. Montreal, Colorado, Philly, Ana, NYR. rise and fall |
|
Talkshowhost
Ottawa Senators |
|
Location: Ottawa, ON Joined: 10.17.2010
|
|
|
These are the same question:
The Top Teams Are Getting Better and the Lesser Teams are Getting Worse
The Lesser Teams are Getting Worse but The Top Teams are Getting Better
|
|
|
|
With the current point system (i.e. one point for a OTL/SOL), it takes 73-games for skill/talent to overwhelm luck in the standings.
Var(talent) = 12.3 squared minus 8.44 squared
= 8.95 points squared.
So, over an 82-game NHL season, talent is only barely more important than luck -- an SD of 8.95, vs. 8.44.
Similar looks have luck influencing about 37% of standings variance.
Considering how little you can discern from single-game outcome (it's almost nothing - iirc it's somewhere like 0.983), I'm guessing the win percentages for thirty teams, about eight-games per falls well within the expected standard deviation here.
|
|
big_dion
Toronto Maple Leafs |
|
|
Location: I've been successful in business for years which is why I can be on hockeybuzz. - HH Joined: 08.23.2010
|
|
|
lil confused here Ek
what i think is happening is the good teams/franchises you have players taking pay cuts to stay on this powerhouse of a team.
500k this guy, 1 mil off this guy = 5 mil in the end for another all star.
pitsburgh is a prime example, detroit was like this for years as well. |
|
Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
With the current point system (i.e. one point for a OTL/SOL), it takes 73-games for skill/talent to overwhelm luck in the standings.
Similar looks have luck influencing about 37% of standings variance.
Considering how little you can discern from single-game outcome (it's almost nothing - iirc it's somewhere like 0.983), I'm guessing the win percentages for thirty teams, about eight-games per falls well within the expected standard deviation here. - Travis Yost
yes, I know. however, I am just looking at start of the seasons...roughly first 8 games or so per team. |
|
Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
These are the same question:
The Top Teams Are Getting Better and the Lesser Teams are Getting Worse
The Lesser Teams are Getting Worse but The Top Teams are Getting Better
- Talkshowhost
fixed
|
|
whipper334
Calgary Flames |
|
|
Location: The man they call Reveen!! Joined: 01.06.2010
|
|
|
KJohnson
Vancouver Canucks |
|
Location: Nanaimo, BC Joined: 01.16.2013
|
|
|
Interesting thoughts. It's worth noting that it's truly the last 4 years where the differentials seem to be on the rise, and around 2008-2009 and the couple of years previous it was closer to what it was now. What do the full season #s look like?
Tough to draw any conclusions on this small of a sample size. Two of the examples mentioned from both ends of the spectrum, Colorado and New York, have numbers that will assuredly revert back closer to reality. Colorado doesn't have the defense to keep that GAA up, and New York has goaltending and a strong D corp that will definitely bounce back from their current pace. Those were just the obvious examples. I'm pretty sure the Sharks can't keep up that ridiculously high GF they are currently posting, either. They got off to a very hot start to last season 7-8 games in before falling apart for the next 10. You should follow up and revisit at the 41 game mark. |
|
|
|
yes, I know. however, I am just looking at start of the seasons...roughly first 8 games or so per team. - Eklund
Yes, I think that's just where you're getting sample size volatility, all within the expected standard deviation for eight or so games.
It's sort of the equivalent of a sports league where you can flip a slightly-loaded coin one-hundred times, then look at the data for only the first twelve flips.
The coin, loaded slightly favorably to tails (in this case, representing skill), may land 53% of the time over the course of the season. But reduced to twelve games, you're much more likely to see disparate results. |
|
Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
Interesting thoughts. It's worth noting that it's truly the last 4 years where the differentials seem to be on the rise, and around 2008-2009 and the couple of years previous it was closer to what it was now. What do the full season #s look like?
Tough to draw any conclusions on this small of a sample size. Two of the examples mentioned from both ends of the spectrum, Colorado and New York, have numbers that will assuredly revert back closer to reality. Colorado doesn't have the defense to keep that GAA up, and New York has goaltending and a strong D corp that will definitely bounce back from their current pace. Those were just the obvious examples. I'm pretty sure the Sharks can't keep up that ridiculously high GF they are currently posting, either. They got off to a very hot start to last season 7-8 games in before falling apart for the next 10. You should follow up and revisit at the 41 game mark. - KJohnson
I agree completely and fully intend to. thanks! |
|
|
|
Yes, I think that's just where you're getting sample size volatility, all within the expected standard deviation for eight or so games.
It's sort of the equivalent of a sports league where you can flip a slightly-loaded coin one-hundred times, then look at the data for only the first twelve flips.
The coin, loaded slightly favorably to tails (in this case, representing luck), may land 53% of the time over the course of the season. But reduced to twelve games, you're much more likely to see disparate results. - Travis Yost
More to this: we ran parity #s for a dozen or so leagues (JW), and there's fairly regular bounce. Some leagues generally have more parity than others, but the relationship between skill/luck over longer samples (i.e. years) on a per-league basis seems to be flat.
Graph 1 |
|
Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
Yes, I think that's just where you're getting sample size volatility, all within the expected standard deviation for eight or so games.
It's sort of the equivalent of a sports league where you can flip a slightly-loaded coin one-hundred times, then look at the data for only the first twelve flips.
The coin, loaded slightly favorably to tails (in this case, representing luck), may land 53% of the time over the course of the season. But reduced to twelve games, you're much more likely to see disparate results. - Travis Yost
you know stats way better than I do, but there is a definite rise in the differential and without knowing stats at all, just watching several games a night a growing difference in the talent levels from team to team.. Having been watching hockey in the heyday of dynasties in the 70s and 80s to the mind-numbing parity of the 90s through 2009 I can more feel this shift than prove it on a small sample size. |
|
WaterBoy
|
|
|
Location: Gardez-le votre ANGLAIS, YT Joined: 06.27.2006
|
|
|
Title makes me |
|
RealityChecker
Vancouver Canucks |
|
|
Location: I stay away from the completely crazy rumours on the internet.I will occasionally debunk them-Eklund Joined: 04.18.2010
|
|
|
you know stats way better than I do, but there is a definite rise in the differential and without knowing stats at all, just watching several games a night a growing difference in the talent levels from team to team.. Having been watching hockey in the heyday of dynasties in the 70s and 80s to the mind-numbing parity of the 90s through 2009 I can more feel this shift than prove it on a small sample size. - Eklund
bring back the ek factors. |
|
Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
bring back the ek factors. - RealityChecker
I will. thanks for asking |
|
WaterBoy
|
|
|
Location: Gardez-le votre ANGLAIS, YT Joined: 06.27.2006
|
|
|
I will. thanks for asking - Eklund
Could all of this be related to the diminishing cap ? |
|
SPIbanez
Edmonton Oilers |
|
Location: Winnipeg, MB Joined: 08.12.2010
|
|
|
I feel the better teams keep getting better while the bad teams stay the same.
Free agents are going to the better teams.. It seems like for the last 5 years at least, it doesn't really matter as the same teams stay good and the same teams stay bad.... i.e. Detroit.pitt.boston.sanjose.chicago.vancouver... vs EDM,Florida,Buffalo,Tampa, .. It's nice to see teams like colorado and toronto finally climbing the standings . It doesn't hurt to have some variation in the playoffs.. |
|
Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
Could all of this be related to the diminishing cap ? - WaterBoy
GREAT point and question. this year especially. |
|
blizzzard
New Jersey Devils |
|
|
Location: Orillia, ON Joined: 07.02.2011
|
|
|
SotaPopinski
Minnesota Wild |
|
|
Location: Minny Joined: 02.21.2011
|
|
|
It's a phenomenon about as curious as global climate change |
|
SotaPopinski
Minnesota Wild |
|
|
Location: Minny Joined: 02.21.2011
|
|
|
So who is getting traded where ? - blizzzard
Miller, Vanek to Edmonton for Dub and Yak |
|
Eklund
Commissioner |
|
|
Joined: 09.15.2005
|
|
|
So who is getting traded where ? - blizzzard
well you know me....this is all heading that way I assure you! |
|
WaterBoy
|
|
|
Location: Gardez-le votre ANGLAIS, YT Joined: 06.27.2006
|
|
|
well you know me....this is all heading that way I assure you! - Eklund
Off topic. While in MTL.. where did you watch the game from on the plateau ? |
|