How Screwed are the Flames?

[EldrickOnIce]

We go back 10 years with this. From our old friend, Mudcrutch (a pioneer in statistical based prediction).
Would be interested in Delow still saw this as the best predictive statistic.
Possibly the best quote from the article

Wins and losses are what matters, not some equation that manages to offer a guess.

The Value of Outscoring: A Primer

http://oilfans.com/article.php?id=NEWS-2005-08-24

[White Out 403]

I think maybe we're talking about 2 separate things or I'm possibly confused.

Goal differential is a great predictor of where teams are now
Corsi is a good predictor of where teams likely are headed

yes? no?

[EldrickOnIce]

Originally Posted by Resurrection

I think maybe we're talking about 2 separate things or I'm possibly confused.

Goal differential is a great predictor of where teams are now
Corsi is a good predictor of where teams likely are headed

yes? no?

Maybe.
But I expect one can make an argument that teams who outscore are likely to continue to do so... Just as teams who out shoot are likely to be better.
But yeah, if you are always chasing the puck, it's much less likely to continue. But that was the whole crux of the Flames v Oilers this season. The Oilers had to improve dramatically and the Flames had to falter, just as dramatically. And we know how that went.
Of course there are anomalies. And I guess both the Flames and Oilers are. That doesn't discount 'advanced stats' - just shows they simply generally a model of outcomes. I just don't think it's that great a model.

[getbak]

Originally Posted by EldrickOnIce

Goal differential is a better predictor.
Perhaps one team a year with a negative goal differential makes the playoffs. So yes, you can use smaller sample sizes to predict with it, just like you can with small corsi sample sizes.

Goal differential is just far more accurate.

How far in advance is it accurate though?

On December 1, Toronto had a positive goal differential and they were in a playoff spot. Today, Toronto has a negative goal differential and they're well out of a playoff spot. On both dates, they are where they should be based on goal differential, but their December 1 position did nothing to predict their March 8 position.

[White Out 403]

Originally Posted by EldrickOnIce

Maybe.
But I expect one can make an argument that teams who outscore are likely to continue to do so... Just as teams who out shoot are likely to be better.
But yeah, if you are always chasing the puck, it's much less likely to continue. But that was the whole crux of the Flames v Oilers this season. The Oilers had to improve dramatically and the Flames had to falter, just as dramatically. And we know how that went.
Of course there are anomalies. And I guess both the Flames and Oilers are. That doesn't discount 'advanced stats' - just shows they simply generally a model of outcomes. I just don't think it's that great a model.

Shouldn't this be easy to test? take 20 game segments and see how in that segment what predicts better, corsi or GD in that 20 games?

[Street Pharmacist]

Originally Posted by getbak

How far in advance is it accurate though?

On December 1, Toronto had a positive goal differential and they were in a playoff spot. Today, Toronto has a negative goal differential and they're well out of a playoff spot. On both dates, they are where they should be based on goal differential, but their December 1 position did nothing to predict their March 8 position.

Can't find the article now (still searching), but when Fenwick Close, Corsi 5v5 and goal differential are compared at various points through the season goal differential was worst early on, and only slightly less than Fenwick as the season progressed. Fenwick was the best all the way through.

There's lots that doesn't explain. Maybe in general that's true, but what if for example, a team has an elite goal differential but a poor Fenwick close (as in the flames situation)?

[Flashpoint]

Originally Posted by CorsiHockeyLeague

Right now, the Kings have fewer wins than the Canucks, Wild and Flames, but in my opinion at least they're better than all three of those teams.

I was at their 1-0 loss against the Pens last night, and I can assure you they aren't. The only reason people seem to think so is because of them winning it all last summer.

The current edition of the Kings is broken. They've been proving it all season.

[trublmaker]

[QUOTE=CorsiHockeyLeague;5175723] Right now, the Kings have fewer wins than the Canucks, Wild and Flames, but in my opinion at least they're better than all three of those teams.




well you'd be wrong because all 3 of those teams are ahead of the kings at the moment.

[CorsiHockeyLeague]

^See, posts like that are why you get the ascerbic and sarcastic responses from people on twitter.

Originally Posted by Flashpoint

I was at their 1-0 loss against the Pens last night, and I can assure you they aren't. The only reason people seem to think so is because of them winning it all last summer.

The current edition of the Kings is broken. They've been proving it all season.

Yeah, just like they prove it every season, before getting in in the bottom half of the seeding and running everyone's show. People think they're good because they're damned good. The fact that you watched a 1-0 overtime loss against another good team in which LAK walked all over the Pens for large segments of the evening doesn't really mean a whole lot to me.

Any of the teams I mentioned there plays the Kings in a 7 game series (which seems not at all unlikely) I'm picking the Kings every time.

[EldrickOnIce]

Originally Posted by CorsiHockeyLeague

Any of the teams I mentioned there plays the Kings in a 7 game series (which seems not at all unlikely) I'm picking the Kings every time.

Must be a typo, as sample size is key for corsi
Did you mean to say you would take the Kings in a best of 107?

[Resolute 14]

Originally Posted by CorsiHockeyLeague

^See, posts like that are why you get the ascerbic and sarcastic responses from people on twitter.

And comments like this are why the analytics crowd has so much trouble finding respect.

Anyway, for those arguing about LA's home ice bias, the Kings have the best CF% on the road, and their average is only slightly below their home numbers. They are where they are because they struggle badly to score on the road, and because their special teams are mediocre to awful.

[ah123]

For all the fans who are "fretting" about the discrepancy between the Flames' results and their advanced stats, you should read this article Darren Haynes:

http://flamesfrom80feet.blogspot.ca/...-say-five.html

His advice:

Back in the early 80s when I was teenager watching Flames games on TV, I loved it. I couldn't tell you then or now whether they were supposed to be winning or not, whether they were lucky or not, and I didn't care.

If there's a single bit of advice for this season. Adopt that same attitude. Don't over think it, just enjoy it. Man, it's just hockey and it's been fun.

It is a very refreshing approach; enjoy the success now and let the chips fall (next season) where they may, because you have no control over it anyway

[PeteMoss]

Until advanced stats get beyond counting shots (and some do, although a lot of those guys have been hired by NHL teams by now) its not going anywhere. There is approximately 8 billion articles every year of people doing their own Corsi proving analysis and its beyond boring. Time to move beyond it and take the next step.

That said - yes the Flames get outplayed in a lot of games. But I rarely see a Flames goalie playing beyond a normal goalie or the Flames scoring boatloads of awful goals that should be saved.

But really - I don't care one way or another. The Flames can have the worse Corsi in the league but if the games are exciting and the team is building into something, I'm a happy fan.

[troutman]

Stats don't really account for how teams play differently once they are up or down a goal.

[Imported_Aussie]

In looking at "advanced" stats more, I keep coming back to one area where I find them tough to accept - they treat shots the same without any ability to look at shot quality, and they assume many instances in hockey are due to luck, which will regress to a mean over time. Though regression over time would happen, that period of time would be (IMO) longer than a season, maybe even longer than 2 seasons, which makes that assumption tougher to accept. Also, hockey analytics ignore shot quality, even though that has been proven in basketball to be a great way to measure effectiveness - as different players have better chances of scoring from certain positions, and overall trends show certain areas are more likely to produce scoring.

What would be awesome to see if the shot % outliers, then compare the zones they have shot in, and success rate in each place - removing the few empty net goals. That would be a way to track if a shooting percentage was really off, or if a player was changing/forced into changing his game.
Same thing could go for teams that have less shots but score a high percentage (e.g. the Flames) - do these shots come from certain areas, how much in each section of the offensive zone - and how does that compare to effective shooting percentages across the league?
Break the offensive zone into sectors - left and right sides, at the point, hashmarks, crease, etc.

[CorsiHockeyLeague]

I keep coming back to one area where I find them tough to accept - they treat shots the same without any ability to look at shot quality

Someone literally said exactly what you said about 2 pages ago, I responded. That's fine, but does everyone see what I mean about having to deal with the same exact counterarguments over and over again? No criticism to imported_aussie, it just gets frustrating after a while.

Incidentally, to add something else, you can go to war-on-ice and see heat maps that show you where teams are taking most of their shots from, which I think addresses your question.

Originally Posted by EldrickOnIce

Must be a typo, as sample size is key for corsi
Did you mean to say you would take the Kings in a best of 107?

I did mean a best of seven. I'd pick the Kings in a single game, too, but I'd just have a lower chance of winning that bet.

Originally Posted by Resolute 14

And comments like this are why the analytics crowd has so much trouble finding respect.

Please explain why you'd take that view. The statement "You're wrong, clearly X team is better than Y team because they're higher in the standings" barely deserves a response even if you don't believe in any "advanced" stat. It's clearly wrong, and we've just been talking about how obvious that is.

Originally Posted by PeteMoss

Until advanced stats get beyond counting shots (and some do, although a lot of those guys have been hired by NHL teams by now) its not going anywhere. There is approximately 8 billion articles every year of people doing their own Corsi proving analysis and its beyond boring. Time to move beyond it and take the next step.

Read a comment last week from, IIRC, Travis Yost saying he doesn't get this kind of statement, specifically on the basis that almost no one writes articles based solely around corsi anymore. People have taken the next step.

Originally Posted by troutman

Stats don't really account for how teams play differently once they are up or down a goal.

Yes they do. It's called score effects. Any decent stats site (stats.hockeyanalysis.com) will provide you with data for different game states - 5 on 5, 5 on 4, 4 on 5, up 1, down 1, tied, close, etc. This is actually fundamental to using these stats, and a point of contention - at what stage does the sample size get big enough for score close situations that the signal outpaces the noise.

[Enoch Root]

Originally Posted by CorsiHockeyLeague

^See, posts like that are why you get the ascerbic and sarcastic responses from people on twitter.

Yeah, just like they prove it every season, before getting in in the bottom half of the seeding and running everyone's show. People think they're good because they're damned good. The fact that you watched a 1-0 overtime loss against another good team in which LAK walked all over the Pens for large segments of the evening doesn't really mean a whole lot to me.

Any of the teams I mentioned there plays the Kings in a 7 game series (which seems not at all unlikely) I'm picking the Kings every time.

I am curious as to what your point is with the Kings. You think they are the best of the teams in the race. Well no sh1t Sherlock! Hey guess what - the 2-cups-in-the-last-3-years Kings are pretty good!

Are you attempting to suggest that you learned this by following Corsi? Because pretty much everyone in the hockey wold, analytics or no, can see the same thing.

[GranteedEV]

Originally Posted by CorsiHockeyLeague

Any of the teams I mentioned there plays the Kings in a 7 game series (which seems not at all unlikely) I'm picking the Kings every time.

The Kings have to make the playoffs to play those teams in a 7 game series. Which with their brutal schedule isn't a sure thing.

And don't forget how many years Calgary was regularly near the top of Corsi in the dump n chase era and getting bounced in the first round.

[Red Menace]

It would be nice if these analytic gurus would give some credit for teams like the Flames who are defying the odds, rather than merely dismiss them as being lucky.
That is what is so incredibly annoying for us fans.
Of course we are not one of the elite teams in the league, but we have gotten our place in the standings from hard work, and yes talent, not just because the bounces have gone our way.

[CorsiHockeyLeague]

Originally Posted by Enoch Root

I am curious as to what your point is with the Kings. You think they are the best of the teams in the race. Well no sh1t Sherlock! Hey guess what - the 2-cups-in-the-last-3-years Kings are pretty good! Are you attempting to suggest that you learned this by following Corsi? Because pretty much everyone in the hockey wold, analytics or no, can see the same thing.

It's weird how people get so up in arms about advanced stats people being aggressive and assertive, but this sort of post is just fine. I'm trying to be civil here specifically because people will jump all over me if I don't exhibit the utmost in politeness (see Resolute's last post). Weird double-standard and I'd point to it for the next person who takes the view like MMF and others have (yourself included) that people who write about these metrics are smarmy and arrogant.

Anyway, as to the quality of the Kings, they couldn't necessarily see it in 2012 before their first cup, but even Canucks bloggers who used these numbers were saying their president's trophy winning team was going to lose in rd 1. I'd again point to the above post by "trublmaker" as an extreme example of people coming to the conclusion that they're suddenly not any good anymore. Yeah, no, they still are.

Originally Posted by GranteedEV

The Kings have to make the playoffs to play those teams in a 7 game series. Which with their brutal schedule isn't a sure thing.

True. I strongly believe they will make the playoffs. I could be wrong (see my signature), but if I had to make a straight up or down bet, "will they or won't they", I'm betting they do. If I'm right, and they play one of the Flames of Canucks in round 1 (they won't play the Jets obviously), and I have to make a bet as to who will win the series, I'll bet on the Kings. I could again end up being wrong (see again my signature), but that doesn't make the bet unsound.

Anyway, this has all been exhausting and utterly unrewarding (as always with these discussions), so if people want to return to analytics-bashing, fill your boots.