Recent homestand and advanced stats

[Bingo]

interesting small sample size for eye test vs advanced stats.

Phoenix - win
Colorado - OT Win
San Jose - loss

All stats aside I felt the Flames were sloppy against Arizona, lucky to win against Colorado, and unlucky to lose against San Jose.

If I was to blindly rank the games by possession based on what I'm seeing it would be

1. San Jose
2. Phoenix
3. Colorado

Yet the Arizona game was the Flame's 4th worst game of the season for Corsi, yet the Colorado game was the team's 2nd best, trailing only the win in Nashville. (San Jose game was 6th, and closer to where I would have guessed)

Fenwick has the three games ranked similarly but not as extreme.

was my "view" of these games that off?

[Cleveland Steam Whistle]

Nope, not according to my "view" either. I would have ranked them exactly as you did. SJ being the best effort, with a loss coming at the hands of some unforced defensive errors and a lack of puck luck. The Phoenix game being a non pretty, but solid Flames effort against a stingy team. The Colorado win, although not a dominating performance from the Avs, I would have said was the first game in a while where the Flames won but didn't deserve.

You eyeball snap shot their just firms up my thought that while I'm not anti Advanced Stats, one needs to recognize in hockey they are still just another data point for directional analysis. Still huge room for error in them in regards to hockey, and nothing can really beat what a knowledgable fan can see from just watching the game. Which is also why I'm not sweating the advanced stats naysayers this year on the Flames, I am nervous about whether the Flames can keep this type of play up, but not because I think they are getting lucky, as I think to date they have been pretty full marks for their record.

[ComixZone]

My thoughts are in line with yours. SJ was our best effort.

[Resolute 14]

The Flames actually were positive in terms of Corsi in both the first and third periods against Arizona. But the Coyotes were generally trying to come from behind in the second and were pressing for a tying goal. In the second, there was three clear spots where the Coyotes piled up a series of shot attempts in a short time period - and I'm betting those coincide with times Hartley threw the third pair D out against the Coyotes' top line. I recall commenting in the PGT about that and how awful that match-up was for us. Those brief periods slanted a game that was otherwise fairly even in terms of shot attempts.

Colorado I felt was pretty even, back and forth game all around, which the stats agree with.

The Sharks game was definitely one of our better efforts. Though the Corsi chart for all situations certainly reveals how much impact the referees had on the last five minutes.

[Bingo]

here are the Flames best and worst games of the season according to advanced stats.

Corsi
1 NSH Oct 14
2 S.J Dec 06
3 OTT Nov 15
4 MTL Oct 28
5 CAR Oct 23

Surprised that road game in Nashville tops the game in Montreal or home to Carolina. Ottawa game felt dominant, and the Flames lost their 2nd best Corsi game (happens).

28 CHI Oct 15
27 EDM Oct 09
26 NSH Oct 31
25 ARI Dec 02
24 S.J Nov 26

Nobody would argue with the game in Chicago. The game in Edmonton could be an example of a team shooting from everywhere, as I never felt in danger in that one. The next two are the Arizona game just played and the win in San Jose, both felt like solid performances from the Flames (Arizona sloppy)

Fenwick
1 MTL Oct 28
2 OTT Nov 15
3 CAR Oct 23
4 MTL Nov 02
5 NSH Oct 14

Similar list, different order.

28 CHI Oct 15
27 EDM Oct 09
26 CHI Nov 20
25 T.B Nov 06
24 S.J Nov 26

This one has that entertaining loss to the Hawks on home ice, Flames played great in that game.

[Finger Cookin]

[Flames Fan, Ph.D.]

I would concur with your view.

As for the seeming "discrepancy" between the eye test and the stats test, I would say that's not too unusual when looking at measures that have demonstrably poor correlations to wins.

[wireframe]

Your view seems right to me. Corsi and Fenwick are stupid stats that are billed as 'advanced' only because they are newer than +/-. Real possession numbers will be available in the NHL eventually, but for now we have to deal with these weak approximations. There is a reason that NHL teams keep their analytics departments secret.

[Freeway]

Big issue was that the team was fairly inconsistent in the San Jose and Colorado games, while pretty even-keeled throughout the Arizona game.

[blankall]

Fenwick is a better stat for the Flames as it takes into account blocked shots. That being said advanced stats are garbage. I'd rather have a team that can put the puck in the net than one that's great at cycling around the perimeter.

[Street Pharmacist]

It's a bit tough. The "eyeball" test is heavily flawed because it's due to memory. Your going to remember certain parts more vividly and weigh them differently. For example, if the team gets outplayed wildly in the first, a tie in the second, and slightly outplays the opposition in the third, it'll feel a lot more more a good solid victory. One could argue that for the game as a whole they probably were lucky not to have been down early.

[edslunch]

Originally Posted by blankall

Fenwick is a better stat for the Flames as it takes into account blocked shots. That being said advanced stats are garbage. I'd rather have a team that can put the puck in the net than one that's great at cycling around the perimeter.

Can someone explain why one would use Fenwick instead of Corsi and vice versa? If the idea is that shots equate possession (you have to possess the puck in order to shoot it) then why not count everything? But then I hear people say that blocked shots shouldn't count since it is a skill and not random (???) and correlates better with wins (marginally). And I see lots of discussions where both are used seemingly in the hope that at least one of them fits.

It seems like we should pick one and stick with it, otherwise we're cherry picking?

[D-Red]

Maybe someone who has a better knowledge can help me out here... Is there any advanced stat that takes into account "scoring chances" or weighting the shots based on a shooting zone. I feel like using those as a factor would help in providing a better representation of sustainability.

I look at a game like the one vs the Oilers where I imagine our PDO was brutal as well as our corsi, yet the majority of their shots came from the perimeter. Only about 7 of their 40 shots were in a prime scoring area, where Calgary was at about 10 of their 26. It's not really "luck" that we were able to capitalize on actual chances as opposed to perimeter shots an NHL caliber goalie should be saving anyway...

[driveway]

Our fan base sounds exactly like the leafs, avs, and wild fa bases of past years.

Whether or not corsi and fenwick measure what they claim to measure is irrelevant. They correlate strongly with winning, and it is a bad sign that we suck at them.

[Iowa_Flames_Fan]

Originally Posted by driveway

Our fan base sounds exactly like the leafs, avs, and wild fa bases of past years.

Whether or not corsi and fenwick measure what they claim to measure is irrelevant. They correlate strongly with winning, and it is a bad sign that we suck at them.

I think your comment reveals the exact problem with "advanced stats," which isn't the stats themselves but how they are being used.

Strong Corsi and Fenwick says may correlate with winning, but the second part of your post was that THEREFORE "it's a bad sign that we suck at them." Being tall correlates with being rich, but that doesn't mean that "it's a bad sign that Donald Trump isn't tall."

All that these advanced stats can tell us is that good teams tend to score better on those metrics. But here's the thing: any number of "non-advanced" stats will give us the same information: good teams tend to have good goal differential, take more shots on goal, have better power play numbers and so on and so forth. Yet no one hails any of those metrics as the "thing teams must do if they want to succeed." We all recognize that these are simply one variable in a complex matrix of variables, and occasionally a team succeeds despite not doing well at one or more of them, or fails despite looking good on most of those measures.

Frankly, statistical analysis in hockey is in its infancy, and we should be cautious about raising it on too high a pedestal this early. One of the things we need our stats gurus to develop is a healthy skepticism about what these metrics are actually telling us, so we don't jump to the conclusion that a stat which correlates with success can also predict success for a given team, because logically that just doesn't follow.

[driveway]

I don't think that the Flames must rate high in these metrics in order to succeed, I'm saying I would be a hell of a lot more comfortable saying we were going to keep succeeding if we were rated highly.

Yes, there are such things as statistical outliers, statistics are only probabilistic, etc. etc. However, we haven't yet seen a team be as poor as we are in these metrics and maintain success. And I also don't believe that the Flames have discovered a new way to play hockey which "beats" these metrics. Something I'm sure we remember from the Leafs and their "shot quality, not quantity" talk.

We have some things going for us, specifically Giordano and Brodie that could make us the outlier.

I just want to see us doing well in every way it's possible to measure hockey, in order to feel confident in success going forwards.

edit: Also, you know full well there is a difference between a correlation and a strong correlation. Corsi and Fenwick scores have a much better correlation with winning than height with wealth, that's a pretty poor argument.

[PeteMoss]

They don't have a particularly high correlation to winning. They just correlate to winning more than anything else anyone has found at this point.

[Bingo]

Last night's game in Toronto comes in as the 6th best Fenwick game this year for the Flames, just ahead of the previous loss to San Jose.

They were winning with bad numbers, and now losing with good numbers.

Overall probably a good sign that the team is
a) playing better and better
b) not getting as lucky

[Bingo]

oh and last night's game ranks #1 for the Flames in Corsi

First time over 60% at .6039

Updated top 5 and bottom 5s, watching for more recent games in the top, and more distant games in the bottom

Corsi
1 TOR Dec 09
2 NSH Oct 14
3 S.J Dec 06
4 OTT Nov 15
5 MTL Oct 28

28 EDM Oct 09
27 NSH Oct 31
26 ARI Dec 02
25 S.J Nov 26
24 ANA Nov 18

Fenwick
1 MTL Oct 28
2 OTT Nov 15
3 CAR Oct 23
4 MTL Nov 02
5 NSH Oct 14

28 EDM Oct 09
27 CHI Nov 20
26 T.B Nov 06
25 S.J Nov 26
24 FLA Nov 08

[edslunch]

Originally Posted by PeteMoss

They don't have a particularly high correlation to winning. They just correlate to winning more than anything else anyone has found at this point.

Perfect correlation is r-squared = 1.0. The various flavours of Corsi and Fenwick are between 0.25 and 0.3 - correlated but not what I would term strongly correlated.