Unsustainable 2014-2015

[DL44]

For people here who follow the numbers...

In the same vein as how many were predicting the potential demise of the Avs and Leafs last year... i shall ask now... at this early stage... Are there any outliers thus far where On Paper the numbers don't appear to support the results?

You gets bits and pieces in various threads...
Like I've read... Chicago's slow start doesn't, so don't be surprised when they win 7 straight shortly... for example.
or..
Avs are playing to what their numbers are...

So generally speaking...


What teams so far are unsustainability disappointing? i.e. i read Chicago somewhere
What teams so far are unsustainability great?
Or teams in between... avg but should better or worse than they are..
(but i would guess advanced stat would be only most relevant for the more extreme teams of the spectrum)


(and then on a side, or maybe they're an example... I'm curious what the numbers say about the Canucks?)

Click to read more...

 

[Bps21*]

Minnesota is playing a lot better than their record and will be just fine. Chicago is too.

Buffalo is just as bad as they look. Even with the injuries Columbus is way outplaying them and are actually underperforming even with the setbacks.

 

[TheKrebsCycle]

Habs no hate. They are barely a positive goal differential team and the record truly flatters . Price is amazing tho so there's that...

 

[charlie1]

The Islanders are a funny case because a lot of their stats were unsustainable but in opposite directions. Their Shooting % was ridiculously high early on, but their Save % was extremely low. Both of those things have since regressed to the mean but they're win % has remained about constant throughout (which makes sense).

 

[Xelebes]

The losing streak by Edmonton against the West is unsustainable. We just have to, I mean, just have to get a win against them sometime.

 

[West]

Crazy High SH% with lots of shots
Corey Perry
Chris Kunitz
Gustav Nyquist
Nick Foligno
Marcus Johansson
Rick Nash
Brock Nelson
Tyler Seguin
Derick Brassard
Tyler Toffoli

Defense
Dennis Wideman


Crazy Low SH% with lots of shots
Chris Stewart
Jakob Silfverberg
David Perron
Andrew Cogliano
Matt Nieto
Emerson Etem
Cal Clutterbuck
Matt Moulson
Ryan Strome
Nathan Gerbe

Defense
James Wisniewski
Zach Bogosian
Keith Yandle
Kevin Bieksa
Drew Doughty
Ryan Ellis

 

[BB1133]

What teams so far are unsustainability disappointing? i.e. i read Chicago somewhere
What teams so far are unsustainability great?
Or teams in between... avg but should better or worse than they are..
(but i would guess advanced stat would be only most relevant for the more extreme teams of the spectrum)

(and then on a side, or maybe they're an example... I'm curious what the numbers say about the Canucks?)

I'll take a shot at this.

Based on "analytics" ...

Teams that are not as good as their record.

Montreal
Toronto
Ottawa
Vancouver
Calgary
Anaheim

And yes it's just a coincidence they are mostly Canadian teams. A lot of them are getting bailed out by elite goaltending, but that will come back down to earth as the season goes on. As far as Vancouver is concerned, they are just an average possession team, with above average shooting % and below average save %. Some of that is because of Lack dragging the numbers down, so they should be able to sustain as a good team the rest of the way. Likely as a wild card team.

Teams that aren't as bad as their record indicates...

Edmonton
Carolina

Teams that should have a better than average record..

Washington
Minnestoa

EDM, CAR, MIN, and WSH need better goaltending because all the other metrics look good. The Wild should be one of the best teams in the league given how good they are at controlling the game.

Teams that should be worse than average

Philadelphia

Flyers are getting carried by their shooting % for the most part. Normally that's unsustainable, but not impossible within one season. I highly doubt that Mason can continue to put up good numbers given how poor they are as a possession team.

Everyone else falls somewhere in the middle or are where they should be in the standings. At around the 20 game mark we'll have a really good idea of what most teams will be the rest of the season.

 

[DL44]

I did Vancouver at the 10 game mark... (7 games ago)...

And their numbers all supported their 7-3 record and how well they played (i think)..

Their numbers now have definitely fallen back with the California road trip : Out shot 19-37 on back to back nights will wreck a corsi and fenwick nicely..... including a 23-78 corsi thrashing by the Sharks.
but hey... 1-1 in those 2 games.


Some notable stats at the 10 game mark...

10 gms: 7-3-0, 14 pts, 0.700, 2nd in Pac, 3rd in West, 5th Overall

Goals/gm
10 gms: 3.30, 5th

GA/gm
10 gms: 2.90, 20th

Goal Differential
10 gms: +5, 8th

PP %
10 gms: 23.7, 7th

PPG
10 gms: 9, 4th

PK %
10 gms: 87.2, 5th

PPGA
10 gms: 5, 6th

Fenwick Close
10 gms: 53.7, 8th

Corsi Close
10 gms: 53.6, 6th

Shots on Goal
10 gms: 32.5, 4th

Shots on Goal Against
10 gms: 29.1, 13th

So the numbers seemed to be in line for their 7-3-0 start... but everything seems to have fallen off stats-wise over the 7 games since... except the wins.. 5-2.
So i can see how the numbers will show they're performing better than they should be lately..

 

[UnrefinedCrude]

Teams that aren't as bad as their record indicates...

Edmonton
Carolina

This points out a flaw in just using analytics here.

The Oilers are losing in big part due to "the big mistake" that leads to a killer scoring chance and/or goal.

So really they aren't playing better than their record. But there is no real metric that tracks lapses and breakdowns that lead to goals. But they aren't just random or fluke events, they are a problem with the team, not a statistical anomaly.

 

[mkwong268]

All teams give up goals from big mistakes you just don't remember or notice them as much when you aren't as invested in the team. Even the great Jonathan Quick mishandles the puck that costs LA goals.

Our poor shooting percentage is costing us as much as our save percentage which includes those goals given up from big mistakes. Both are down about a percent from last year in 5v5 situation and are unlikely to stay down.

 

[ottawa]

Habs are as good as their record suggests, they just weren't playing up to par with is until recently. Look at their past 6 games, the goalie didn't have to steal any one of those wins because the players that had a slow start are finally starting to play as they should be.

As for goal differential...they outscored their opponents 24-7 during their win streak, that goal differential will slowly get better. You guys are giving it tremendous weight, much more importance than needed.


Their next 4 games will be a real test though...PIT, STL, BOS and NYR

 

[Baemon Severson*]

Habs are as good as their record suggests, they just weren't playing up to par with is until recently. Look at their past 6 games, the goalie didn't have to steal any one of those wins because the players that had a slow start are finally starting to play as they should be.

As for goal differential...they outscored their opponents 24-7 during their win streak, that goal differential will slowly get better. You guys are giving it tremendous weight, much more importance than needed.


Their next 4 games will be a real test though...PIT, STL, BOS and NYR

Yeah, I'm not buying Montreal as a top team yet. 22nd in Fenwick close

 

[Plekaturtlenec]

Yeah, I'm not buying Montreal as a top team yet. 22nd in Fenwick close

They were pretty terrible early in the season despite winning games but have gotten better as of late. This week will be a real test, some very good teams on the horizon.

 

[Houdas]

I made one analysis about this topic based on last 5 seasons. I've found out that the goal differential can actually predict points.

I've used analysis of variance and linear regression.

For example Montreal will have 107 points at the end of the season if my predictions are not really that inaccurate.

 

[DL44]

I made one analysis about this topic based on last 5 seasons. I've found out that the goal differential can actually predict points.

I've used analysis of variance and linear regression.

For example Montreal will have 107 points at the end of the season if my predictions are not really that inaccurate.

Could you just post the final standings based on your analysis?

Would be neat to see how accurate it is at the end of the year... It would also give some context to the Habs 107 pts.

Please and Thank you!

 

[hatterson]

I made one analysis about this topic based on last 5 seasons. I've found out that the goal differential can actually predict points.

I've used analysis of variance and linear regression.

For example Montreal will have 107 points at the end of the season if my predictions are not really that inaccurate.

Goal differential does have a strong correlation with winning over the long run, which should seem obvious given that the way to win is by scoring more goals than your opponent. Based on past analysis I've seen the break even point between shots based metrics and goal based metrics is anywhere in the 60-100 game mark

 

[Houdas]

You can check my blog with commentary and description of my methodology or you can download this excel file where are the results.

Hope there are no mistakes

 

[Feed Me A Stray Cat]

Calgary is in over their heads. 2nd best ES shooting percentage despite having pedestrian talent on their team. Bad possession metrics.

 

[Volodya Krutov]

You can check my blog with commentary and description of my methodology or you can download this excel file where are the results.

Hope there are no mistakes

It would be interesting if you could factor current teams S% and Sv% in your equation. Some teams PDO look unsustainable but still had a direct impact on Goal differential over that small sample size.

 

[Mad Dog Tannen]

I made one analysis about this topic based on last 5 seasons. I've found out that the goal differential can actually predict points.

I've used analysis of variance and linear regression.

For example Montreal will have 107 points at the end of the season if my predictions are not really that inaccurate.

Is goal differential a predictor or a symptom of winning?

Did you use month end data throughout the past 5 seasons to predict final points? How predictive we're October month end, November month end, etc when predicting the final points for each individual season. How much did the predictive ness of goal diff fluctuate from season to season.

Also, what other predictive variables did you compare goal differential to. Is it any more predictive than save percentage, jersey colour, etc.

My concern if your using the final standings for all five years is that it's pretty hard to win games and not score more goals that you allow over 82 games.

Additionally, the points Montreal has and Nashville has in your spreadsheet seem curious given the difference in their goal diff.

I see you removed shootout goals against, did you also remove shootout goals for?

Edit: how did u get the goal diff of +6 for Montreal. I calculate 0, +4 minus the four shootout goals gives u zero.

 

[Houdas]

I've used numbers from official NHL site - here

That +/- stat is without PPF and PPA, so it, as I believe, counts only ES goals.

Those past 5 seasons +/- is calculated on the same basis.

This idea of correlation of Points and +/- came to my mind quite a time ago. I wanted to know, if there is any dependability.
Those predicted points also aren't really that much accurate. It's not 99% accurate, not even 95%. It can be like 70% accurate. I haven't made specific confidence intervals as it is pretty time consuming thing.

You also shouldn't think that this is really a prediction - it's function is more descriptive. You can se, which team is overachieving or underachieving based on past five seasons stats.

Also, Strongest relation between these two stats was in the last season, while in 2011/2012 season the relation was the weakest in the span of past 5 seasons. But thanks to analysis of variance I've found out, that last five seasons correlation of points and +/- isn't really different.

I haven't compared it to anything so far, when it comes to predicting anything. In one of my recent analyses I took teams' sv% and points, it's also on my blog - take a look. But I plan to do some kind of advanced model in the future, I just hope I will have enough time for it. Now I focus on finding some interesing factors, factors that are difference-makers in regular season, like CF%, that I find very important, too, imho.

 

[DL44]

So a bump...

Halfway thru the season... so a pretty significant sample size where a lot of outliers would be hammered out by now and their records would more justified by the numbers....

But are there any teams at the moment with points totals greater than 'the numbers' suggest...
or visa versa...
Teams that have depressed total relative to what 'the numbers' suggest...

 

[InjuredChoker]

oilers and minny are worse than what their numbers suggest but brutal goaltending can do that. if dubnyk is even half have decent, they should be trending up. could see canes improving now that they've been getting healthy.

calgary should fall out of playoffs imo.

habs have some questionmarks on them but price is very good goalie and should carry them to playoffs.

anaheim and nashville have won lot of 1-goal games, which is a red flag. but they should still be good enough to make the postseason.

 

[Vesa Awesaka]

Ottawa has not been as bad as their record indicates since firing MacLean. Everything besides save percentage and PP percentage across the board is better.

 

[Raccoon Jesus]

Kings should be doing better as per plenty of stats but there are also very visible indicators of why they are not. Though if some of those simple things were sorted out, you could theoretically expect a hot streak in the back end of the season.