Official NHL Season 2016-2017 Thread

[Peter_Doherty]

CA/60

Is this Corsi Against per 60 minutes?

A WAR model like DTMAboutHeart

What's this?

Yes, it's Corsi Against per 60 minutes.

Talking about @DTMAboutHeart on twitter who's made his own WAR model, he did a write up on it at seasons start if you are interested, if you can't find it on google let me know and i'll dig it up.

[Peter_Doherty]

One more thing on GF% is that it should mostly be used along with xGF% (expected GF%), PDO and xPDO to get a fair picture since it doesn't account for on ice SH% and on ice SV%.

[nino33]

Talking about @DTMAboutHeart on twitter who's made his own WAR model, he did a write up on it at seasons start if you are interested, if you can't find it on google let me know and i'll dig it up.

Thanks! I found explanations on google https://www.google.ca/#q=%40DTMAboutHeart+war
Looks like I've got some reading to do!


Thanks again Peter!

[Peter_Doherty]

Ur welcome

I'm still not completely sold on that model but it's still an interesting read.

[A9L3E]

Am I the only one here who has no idea about what Peter tries to explain?

[philou21]

No me either. I prefer to watch hockey than analyzing it that much.

Galchenyuk will be out 6 to 8 weeks. That's a huge loss for MTL. Desharnais is out as well for the same time. He's having an horrible season but it's still 2 centers out of the game.

It's going to be interesting to see what Bergevin's going to do about it. He doesn't have the choice to trade.

[CJ]

Am I the only one here who has no idea about what Peter tries to explain?

I understand most of it, but I'm not into analytics.

More to hockey than just possession and shots.

[nino33]

Am I the only one here who has no idea about what Peter tries to explain?

I like to learn, so I'm trying to understand HaHa

It's just "one more tool in the toolbox" to me, one more thing to discuss/debate about, one more thing that the pros use in determining success, one more thing that can be used when editing EHM (or I suppose technically it's more than one as there's multiple advanced stats).....for me they're not necessarily "a big deal" though (as many of the pros believe too), as the bottom line is winning and not any specific stat or stats (and there's outliers with all types of stats, including advanced...where teams that should be good/winning aren't, and teams that shouldn't be good/winning are) & of course there are aspects to a "winner" that don't easily show up statistically

[philou21]

I've heard that the Golden Knights might need to change their logo because it wasn't approved. There is alot of similarity with an highschool team logo.

[Peter_Doherty]

I just really like numbers, always have, it's not for everyone though and that's fine. I really enjoy that you can use the stats to see which teams are sustainable etc

I really hate how black/white it's become though, either you're into analytics or you HATE them (not everyone ofcourse but it's become a hostile talking point).
I don't think this is necessary and both 'sides' are at fault here. Analytics is a great tool since it fills in the gaps in your memory, that's how i use it atleast.
Also it's good for keeping up with teams/players you don't watch that much, even though it won't paint the full picture it gives you a good idea of how they're doing.

nino - "as the bottom line is winning", I agree with this but fact is that 'analytics' is proven to be very predictive on a team basis. Good Corsi has a very high correlation with long term success.

But yeah, if you don't care about the numbers and just want to enjoy watching hockey then feel free, it's not everyones cup of tea.

If anyone want to get into analytics or just have some basic questions about it then feel free to ask, i'd be happy to point you in the right direction or answer your questions.

[Aladyyn]

It's also important to note that these "fancy stats" aren't all that fancy. Having good Corsi = outshooting your opponent. It just has a weird name so people don't trust it.

[nino33]

I guess I don't think analytics should ever be used "alone" in a debate/discussion, as the lack of context puts to much "responsibility" on the analytic stat and I don't think they're predictive enough to support that level of responsibility...in 2014-15 the best Corsi team in the NHL (Kings) missed the playoffs, while one of the worst (Flames) made the playoffs http://www.thehockeynews.com/news/artic ... rst-get-in - that right there (and other similar examples) are why I think the way I do

Last night the Leafs "out attempted" the Wild 87-41 (more than double!) but lost the game (and didn't even get a "loser point") http://www.msn.com/en-ca/sports/news/ma ... ailsignout

From what I've read it seems a big part of the hostility against analytics is the lack of context and additional thought (i.e. traditional hockey talk/thinking) that to many that support analytics don't include

IMO in an actual discussion/debate about specific teams/players I don't believe analytics alone can support/predict anything, because the proof of outliers means that more than analytics is needed.....but they're certainly a useful tool or set of tools to have in the debate/discussion toolbox!

[CJ]

I've heard that the Golden Knights might need to change their logo because it wasn't approved. There is alot of similarity with an highschool team logo.

The problem is the name. There already is "Albany St. Rosen Golden Knights".

[philou21]

Yeah you're right. The article I read got edited now. They were talking about the logo first.

Personally I don't think it's a big deal. Especially since it's high school level and not another pro team. With so many teams out there the possibilities are running short.

[Peter_Doherty]

Sure nino, context is always needed but you can add context by not using only one metric. For example using a combination of Corsi and puck luck (PDO). Fact is though that Corsi in itself is a very good predictor of success long term and i'll 'prove' it to you here.

I'll first show their rank in raw CF% then in () i'll show their rank after adjusting for score, zone and venue.

2016 SCF: Penguins 2nd (3rd), 1st after coaching change. Sharks 9th (8th).
2015: Hawks 2nd (5th). Lightning 4th (4th).
2014: Kings 1st (1st). Rangers 8th (8th)
2013: Hawks 4th (2nd). Bruins 3rd (3rd)
2012: Kings 2nd (1st). Devils 13th (12th)
2011: Bruins: 14th (10th). Canucks 5th (5th)
2010: Hawks 1st (1st). Flyers 14th (16th)

As you can see the only Stanley Cup winner in this time frame to be ranked worse then 5th in either undadjusted or adjusted CF% are the Bruins and they were winning off the back of an incredible Tim Thomas. Not a single of these teams were under 50 CF%. Corsi isn't end all be all but it's VERY predictive of future success.

Here are the teams with the highest CF% during this time span (09/10 - 15/16:

Kings 54.51% (54.46%)
Hawks 53.67% (53.78%)
Red Wings 53.18% (52.85%)
Bruins 52.23% (52.63%)
Blues 52.16% (52.53%
Sharks 52.08% (52.37%)
Penguins 51.92% (52.21%)

So yeah, one final time, Corsi isn't perfect but it's a very good predictor.
All the stats from Corsica.Hockey

[ClassicSwarley]

Just to make sure, if CF% is under 50%, is that below-average/bad?

[Peter_Doherty]

Yes, 50% is average.

[nino33]

Sure nino, context is always needed but you can add context by not using only one metric. For example using a combination of Corsi and puck luck (PDO).

I believe there's more to it than math, more to it than just analytics, that's all

Here are the teams with the highest CF% during this time span (09/10 - 15/16:

Kings 54.51% (54.46%)
Hawks 53.67% (53.78%)
Red Wings 53.18% (52.85%)
Bruins 52.23% (52.63%)
Blues 52.16% (52.53%
Sharks 52.08% (52.37%)
Penguins 51.92% (52.21%)

So yeah, one final time, Corsi isn't perfect but it's a very good predictor.

All teams looking to win the Cup (and not just "make the playoffs"), of which 4 of 7 actually did so (so in terms of what matters, it's not "very good" IMO for what matters)

The Blues haven't even made a Final, so being 5th means very little
The Bruins are not considered a Cup contender anymore, nor the Red Wings, so predicting "future success" isn't "very good" IMO either
And it doesn't mention the 5 other teams that made the Final during the same time period (just as often as Detroit/San Jose)
And as mentioned previously, the best Corsi team in 2014-15 didn't even make the playoffs


Like I've said, analytics are good tools, but there's more to it than just analytics in the opinion of many (me being one of them! HaHa)

P.S. I'd like to repeat though, I do appreciate/value analytics, and I also appreciate you pointing out "the analytics side" with your posts & using your knowledge in providing answers to the questions about analytics when asked. You can have a last word if you'd like, but I don't want it to turn into a conflict/squabble (it's been nice talking and having it be positive!), so I'm going to let it go now

[Peter_Doherty]

No conflict here, maybe my tone was more harsh then i wanted it to be. I was just pointing out that good Corsi teams generally do very well historically.
I agree that numbers don't tell the full story, my view is that they can get very close though if you apply them 'correctly'.
Kings missing playoffs 14/15 comes down to bad special teams, bad luck and just some bad timing, it's an anomaly imo.

[nino33]

No conflict here, maybe my tone was more harsh then i wanted it to be.

Your tone was fine this conversation/discussion's been great

[Peter_Doherty]

It can be hard to talk about numbers without coming across as arrogant, let me know if i do because that's actually the one thing i hate the most about analytics, how everyone sounds like they know more then anyone else.

[Primis]

I've heard that the Golden Knights might need to change their logo because it wasn't approved. There is alot of similarity with an highschool team logo.

The problem is the name. There already is "Albany St. Rosen Golden Knights".

Actually, the even larger looming problem is that the U.S. Army's parachute team is called the Golden Knights.

[A9L3E]

So, is Corsi:

a) games when team out-shot the opponent / total games
b) total shots / (total shots + total shots against)?

[Peter_Doherty]

Corsi is just another name for shot attempts which is every shot taken, no matter if they get blocked or miss the net.
So it just shows which teams shot the puck more, it's not a measure for possession but it's a good predictor.

CF% is Shot attempts for divided by shot attempts for + shot attempts against.

[Peter_Doherty]

Someone should rescue John Tavares from the trainwreck that is New York Islanders, they just re-signed Cal Clutterbuck to a $3.5M x 5 year contract, that's incredibly bad imo.