** 13/14 REGULAR SEASON STANDINGS ** [Updated Till: Week 21]

[Nick]

* TBL & TOR only 1 loss teams.
* Top 3 teams, all in same division, Toronto, Boston and Tampa all 360+, LAK @ 357, STL @ 338, PIT @ 336.

Loving our scoring system now.

Both CAM and I win our weeks, he passes me by stomping on Carolina's throat . In a week that if we faced, I beat CAM 9-5-2. Awesome. What a good comparison for determining playoff seating.


--

More weeks won, less weeks lost, more categories won, better in division record, better in conference record, those category tie points! I tell ya.

[Robin Hood]

* TBL & TOR only 1 loss teams.
* Top 3 teams, all in same division, Toronto, Boston and Tampa all 360+, LAK @ 357, STL @ 338, PIT @ 336.

Loving our scoring system now.

Both CAM and I win our weeks, he passes me by stomping on Carolina's throat . In a week that if we faced, I beat CAM 9-5-2. Awesome. What a good comparison for determining playoff seating.


--

More weeks won, less weeks lost, more categories won, better in division record, better in conference record, those category tie points! I tell ya.

Way to be unbiased.

SO losses count for points in the real NHL too and often teams with fewer wins but more OTL finish higher in the standings because of the 1 point.

I am also sure there have been weeks where you won 14-0-2 while CAM faced a tough team and tied 8-8-0 so that point is invalid. Didn't hear you complaining then.

[Shep]

Nobody cares about your team, Nick.

[Robin Hood]

Nobody cares about your team, Nick.

+1. Thank you Shep.

[CAM]

Lol, I won handily against Carolina, yet had my worst statistical week of the season. Go figure

[Nick]

Way to be unbiased.

My opinion on the topic has been the same for 5 1/2 years. Even when it benefited me, I was pointing out the faults.

SO losses count for points in the real NHL too and often teams with fewer wins but more OTL finish higher in the standings because of the 1 point.

Please defend the loser points. That's perhaps the dumbest aspect of the NHL.

I am also sure there have been weeks where you won 14-0-2 while CAM faced a tough team and tied 8-8-0 so that point is invalid. Didn't hear you complaining then.

Than you have selective hearing. I'm been slamming the stupidity of wins being worth different amounts for years. In fact when I squeaked 10 categories off of Mike I commented how dumb it was that despite a super tight week, I gained 12 points on him, much more than the difference between our teams (even in a hot/col match-up).

Our system would be best described as confused.
We ignore how much you win a category by, but put ALL the value in how many categories you won, that isn't compared against the team you actually faced, but rather against the entire league, not that it's a measure what they did relative to you.

[Robin Hood]

SO losses count for points in the real NHL too and often teams with fewer wins but more OTL finish higher in the standings because of the 1 point.

Please defend the loser points. That's perhaps the dumbest aspect of the NHL.

1. It's not a loser point, it's a tie
2. It happens in the NHL all the time and we try to emulate what they do as much as possible

[Nick]

In the NHL system
Bryma has 25 points to Hong's 21, and Shep's 24...
Josh has 14 to PHX and Cliff's 12
I have 29 points to CAM's 27
Chris has 16 to Billy's 14
Berg's with 18 over like 4 teams ranked ahead in our system - because when he lost he lost bad.

[Robin Hood]

In the NHL system
Bryma has 25 points to Hong's 21, and Shep's 24...
Josh has 14 to PHX and Cliff's 12
I have 29 points to CAM's 27
Chris has 16 to Billy's 14
Berg's with 18 over like 4 teams ranked ahead in our system - because when he lost he lost bad.

Please don't call W-L-T the NHL system lol. NHL doesn't have 21 games in a season.

[Nick]

does it have 336 ?

[Robin Hood]

does it have 336 ?

No but 336 is far more likely to produce a result similar to an 82 game season than 21 ever will due to variance. Come on, Nick you know this lol.

Anyways, Shook put it best in that other thread. I get too caught up with you sometimes. If you get the CC to agree with your stance, more power to W-L-T.

[Nick]

Your reference to statistical variance. although a valued conversation, is misplaced, no need to 'tell me I know something' and LOLing about it; either you have never applied statistical measures in an actual application, or you know what you're saying is inaccurate and are intentionally misleading, to what end I have no idea. This isn't good variance which helps a measure, this is error variance that we shouldn't want, and can easily remove.


It's disappointing to me that we continue to use this system. I do not say that in reference to a personal choice like pepsi or coke, we actually use a statistically inferior measure, it fails us in terms of what we are 'saying' it measures. This isn't the argument about ties (we actually don't use the NHL's system fwiw, I dunno how we would, but I'd entertain the idea), this is a discussion about the accuracy of our measure for comparing teams in our standings.

Our 16 category system does a good job of comparing teams head to head. No it's not the most accurate, however it is fair.

Bringing that score between two teams, over to the entire league comparison, is just unjustified. Not even looking at the impact on playoff races or perceived interest added (that's the pepsi/coke stage of the argument, what's more interesting, live playoff races or stealing a point in a week).

[Robin Hood]

this is error variance

What? No it's not. I would love to hear how in the world this is error variance. Your whole argument is based on this one statement.

[Bruyns]

It magnifies big wins and losses, making winning weeks worth different amounts. Why is a 8-7-1 win against a similarly good team worth so much less than a 16-0-0 week against a bottom feeder?

I have no horse in this race and could argue for either side since they both have redeeming qualities.

[Nick]

there is actually a point you made a couple of threads ago regarding what is ' a game' that we are measuring, where we are measuring properly.

however, at no point has our intention ever been to measure ' a game' of hits per week, or 'a game' of goals in a week.

Given that we've always intended for this to be a head to head comparison of two teams across 16 categories, our scoring system of carrying over categories won against a head to head opponent, to represent our points earned relative to the rest of the league is an statistical error.

I know our goal was never to see how many weeks you could win a category for. In an effort to create more 'depth' value and have some variance in how successful teams can be built, we selected a range of catgeories to compare two teams across. Teams that do more categories well than their opponent, we designated as winning that game.

By your definition we've mistakenly measured 16 games a week. Each one considered a head to head battle worthy of comparison across the league (although not direct).

Game 1:
Did you win +/- versus GM 1 ? yes
Hong did versus GM 2
Hong vs you rank= tie , as you both won the first game of the week

Game 2:
Did you win assists versus GM1 ? No
Hong did versus Gm 2
= win for hong, he's now won one more game relative to you, two games over his direct competitor

Game 3:
Did you win blocked shots versus GM1? yes
Hong did versus GM 2
= another game won by each of you, he's still won 1 more game than you.

repeat for all 16 categories which it turns out you believe are games. and we have 336 games a season, each worth 2 points.


if we want 'a game' of hits in a week, why face the same team for all 16 games per week ?

[Shep]

Most people prefer the current system, so I cannot see that changing. We've had it since the leagues inception, and whether some of us prefer WLT, there's no chance it would change barring an overwhelming 80%+ in favour. Whereas I believe it's about 80% in favour of the current scoring system, which I understand.

Lower teams want to show growth and that is much more evident if you start winning 2-3 of the same categories every week (niche categories like Hits, SHTOI, etc). I can only imagine it's more fun for Dallas facing Boston thinking "I wonder if I can take 4-6 categories from Boston" rather than "well here's a loss, who do I face next week?"

[Robin Hood]

It magnifies big wins and losses, making winning weeks worth different amounts. Why is a 8-7-1 win against a similarly good team worth so much less than a 16-0-0 week against a bottom feeder?

See this is one of the misunderstandings about this.

BOS beats CAR 14-0-2
TOR beats LAK 9-5-2

Is TOR's Win worth less than BOS? NO. Why? See below:

When most teams face LAK, they lose handily 10-6-0 minimum. When TOR faces LAK and wins 9-5-2, that's not the difference between 14-0-2 and 9-5-2. It's the difference between what a bad team would normally lose to LAK on.

This is an important point to understand. Under our schedule, each team must play every other team in the conference at least once.

So MIN plays LAK and ties 8-8-0
CGY plays LAK also and loses 0-14-2

The point differential there between MIn and CGY is 16 - 2 = 14 points. That is equal to 7 wins!

All teams have a similarly difficult schedule. The difference between good teams and bad teams is that when they face a tough opponent they don't cream each other.

The difference, thus, should not be measured in absolute terms but more in relative terms. And it reflects in the standings.

there is actually a point you made a couple of threads ago regarding what is ' a game' that we are measuring, where we are measuring properly.

however, at no point has our intention ever been to measure ' a game' of hits per week, or 'a game' of goals in a week.

Given that we've always intended for this to be a head to head comparison of two teams across 16 categories, our scoring system of caring over categories won against a head to head opponent, to represent our points earned relative to the rest of the league is an statistical error.

I know our goal was never to see how many weeks you could win a category for. In an effort to create more 'depth' value and have some variance in how successful teams can be built, we selected a range of catgeories to compare two teams across. Teams that do more categories well than their opponent, we designated as winning that game.

By your definition we've mistakenly measured 16 games a week. Each one considered a head to head battle worthy of comparison across the league (although not direct).

Game 1:
Did you win +/- versus GM 1 ? yes
Hong did versus GM 2
= tie

Game 2:
Did you win assists versus GM1 ? No
Hong did versus Gm 2
= win

Game 3:
Did you win blocked shots versus GM1? yes
Hong did versus GM 2

repeat for all 16 categories which it turns out you believe are games.


if we want 'a game' of hits in a week, why face the same team for all 16 games per week ?

See that's not the intention. Yes we measure "Hits" as a category as a "game". And yes when Hong beats a GM in it, it is a Win.

BUT that is not what we are actually measuring.

What we are measuring is the magnitude by which Hong beats CGY by for example. Let's say that magnitude is 10-6-0. IF you switch to the W-L-T system you make the stats CGY puts up as meaningless and make the magnitude either 100% or 0%. There is no middle ground.

Categories should not be thought of as "games", they should be thought of as "magnitude determinants". TOR isn't 100% better than LAK after beating LAK 9-5-2 this week. 9-5-2 tells the whole story. Still a win --> a big win because other teams would lose 15-0-1 to LAK.

[Tony]

[Sensfanjosh]

Why is my team the bad one in all these examples?.....

[Nick]

You've now totally contradicted yourself.


Either the 'game' is each category and our standings make some sense, or 'a game' is the week across 16 categories and our standing system is full of statistical error.

Why do we care, outside of the head to head match-up, how many categories you win by? why is that more valued than winning a category by a 3:1 margin ?


our comparison of points in the league standings is off.


On a full analysis this would be a time-sequence comparison with controlled but not equal sample group. Any stats person would throw out any result learned from comparing categories won. It's not a valid measure to compare across on a weekly h2h basis.

edit: it also creates a consistency of comparison into the playoffs, where winning the week is how you move on.

This is not some big effort to benefit myself or my team. A WLT system would make our league standings a more fair measure of our head to head style match-ups. In addition to being a more valid and reliable measure, it would also highlight races much clearer (where our system hides them).