Screen Shot 2014-03-20 at 1.43.15 PM

Via Flickr

For a period of time in November, the Flames were awful. Like, just horrible and awful and garbage and every other word that you can think of that meant bad. Their record in October was a perfectly fine 5-5-2,  but the November record of 4-7-2 wasn’t. The question is: why did this happen?

Well, there’s three main water carriers on the team this year: Giordano, Brodie and Backlund. Giordano plays more with Brodie than anyone else (and vice versa) and Backlund has been playing and thriving against top six players for a couple of years now. Even with only three guys to carry the load, the Flames are better than a few teams at even strength. So, did their missing ice time cause the November swoon or was it something else?

First up, we have Giordano. Obviously losing a player as high a calibre as Giordano would have an impact on most of the teams in the league, and it just so happens that Giordano was hurt at basically the exact same time the swoon occurred. Let’s see the splits:

Overall Team CF% Team CF% Pre-Injury Team CF% w/o Giordano Team CF% Post-Injury
46.5% (42 GP) 45.6% (8 GP) 45.5% (18 GP) 47.4%(42 GP)

So, there doesn’t seem to be a meaningful drop from before the injury to the injury in terms of shot attempts. There’s a large improvement afterwards, but I think a large part of that is more regression than anything else. (It’s really hard to be that bad in the NHL, although somehow two teams are below like 44%.) Although I’m fairly certain Giordano had a large impact positively when he came back, I don’t want to mistake causation with correlation – so I’m not going to give him all of the credit yet. It just seems to me that one player can’t have increased the CF% that much by himself.

TJ Brodie was the biggest loser when Giordano went out, as he went from having a line mate who could carry the load to one that was just total suck in Chris Butler. Brodie’s Corsi% suffers a dramatic drop when he is away from Giordano, but it’s hard to say what has more impact on the reduction in his numbers at first glance: loss of a good mate or gain of a bad one.

A cursory look at his WOWYs, however, show that it is the second one. Even though Brodie has played the same type of competition throughout the entire year, his numbers away from Giordano show that at the very least the other defensemen on the team can’t carry the load the way Giordano does. Brodie’s CF% improves when he gets away from basically all of the Flames’ regular defensemen. Funny enough, Giordano and Brodie are almost the exact same when they’re away from each other – 47.2% versus 47.1%. This suggests to me that Brodie and Giordano are highly reliant on each other in that they play like an elite pair when together and are replacement level when apart due to the rest of the team.

Backlund is another case where the timing seems curious. Somewhere around game 40, it appeared – at least by eye – that Backlund was on a sort of TOI upswing. It seemed, to me at least, that maybe the increase in TOI for the best forward on the team was driving the team to better results. Here’s the scatter plot of 5 game samples for Backlund:

Backlund 1

And the polynomial curve:

Backlund 2

Once again, we don’t want to mistake correlation with causation – but knowing what we know about Backlund being an elite player defensively, he’s going to get some credit. The thing that stands out to me is  when he breaks the 30% TOI plane – both the CF% and GF% increase markedly, and at around the same rate. 

On their own, none of these events could’ve had the effect on the Flames that they did. Combined, it did have a little effect. I’m not sure how much, but it wasn’t huge.

Overall Team PDO Team PDO (Pre-Swoon) Team PDO (Swoon) Team PDO (Post-Swoon)
980 (67 GP) 981 (12 GP) 913 (18 GP) 1009 (37 GP)

It turns out that the PDO of the team reveals the answer we seek: in a 12-game sample, a 981 is pretty okay. It’s about average. In any sample, a 913 is garbage. Even if a team is shut out (i.e. SH% of 0), it’s bad – a 913 reflects below-replacement level goaltending. The team was shooting 9.1% over the course of November, which is actually somehow more than average, but only by a tick or so – this year the league’s average SH% is 8.1% and over an 18 game sample that’s within a standard deviation. That means that league average goaltending is about .919. The Flames were not even close to that: .822% over the course of November. Reto Berra is at fault for most of those 18 games. (I still can’t believe that Burke got a 2nd rounder for him. Miracle worker.) I say he’s at fault because that’s 93 points below what replacement level is. In practical terms, that meant an ~5.5 extra goals against, even though the team was scoring at an above-average pace. After the injury, and with more Rämö play, that number has improved to a respectable level even though the team came back to earth in terms of SH%. I can’t understate how negative an effect Berra had on the team.


I believe a variety of institutional factors, bad luck and bad timing were the culprit of the team’s poor November. The combination of increasing Backlund’s TOI, getting Giordano back and getting Brodie away from the black holes that comprise the rest of the Flames’ defense all at the same time, plus an absurdly low PDO turning into a normal one got the Flames out of the tail spin. We know that those three are good players, so them returning should have a positive impact, and we’ve seen that in the results above. I figure that the PDO (i.e. Berra) is at fault for most of the swoon; the only number that appears to either really change significantly in a negative way is the even strength save percentage. If I had to estimate it after looking at the numbers, it’s 40% good players and 60% luck that caused the swoon and the recovery.

So, is it likely to happen again? There were a couple of other rough patches this season, but none as long as this one. I’m going to chalk those up to the normal ebbs and flows that come with a team lacking as much talent as the Flames do. As long as the Flames continue to improve their CF% and get better than replacement-level goaltending (which Rämö has given them this year), I doubt another run like this would be as reflective of the construction of the team.

  • McRib

    Anyone else think that maybe this just isn’t a good team? Based on early predictions, the Flames have overachieved this season. November could have just been a reality check.

    • McRib

      well, in the article i say that it’s really hard to be that bad permanantly. there is too much nhl talent on this team for them to be as bad as buffalo or others simply because of brodie and backlund. people really underrate them.

      • Burnward

        Fair enough. I’ll take your word for it. I’m an Oiler fan, so I don’t see them as much, but I shared season tix to the Flames for years (the last game I went to was Phaneuf’s last as a Flame), so I like to think I know a little about the Flames. Giordano is a great defenseman, but he’s played better this year than others. Brodie’s been good too, and Backlund stayed healhier than last year…I’d say those 3 are the reasons they’ve overachieved. From outside the Flames love-in, I’d suggest that your goaltending, while still not very good, has overachieved this year. So has the temp GM in getting a 2nd rounder for their goalie (MacT paid a 3rd rounder for Scrivens who’s playing great). Monahan has exceeded expectations as a rookie as well. All in all, I just feel like the Flames are overachieving this year.

        • Bean-counting cowboy

          I found it interesting you used the word overachieved a number of times when describing the Flames or one of it’s players. Seriously after watching the Oilers the past 7 years I suppose you would think that hard work, heart, desire and great coaching by a team with half the talent of your first overalls would be overachieving for Oiler fans.

          You have not seen that kind of hockey in a very long time have you??

        • Bean-counting cowboy

          I shared season tix to the Flames for years (the last game I went to was Phaneuf’s last as a Flame), so I like to think I know a little about the Flames.

          erm… how many current Flames players were on that team? Giordano and Glencross.

          Oh, and I guess McGrattan since he was reacquired. I don’t believe Mickis had been called up yet (by the game you saw), and Stajan was only there after the trade.

          re: overachieving. The Flames were projected to be a 30th place team. They are not in 30th place. Therefore, by the definition of the word, they have overachieved.

    • Bean-counting cowboy

      November just compensated for over-achieving at other times. I don’t think this team is as bad as November, or as good as other teams with “relative” winning streaks. Reality is likely somewhere in the middle.

  • mattyc

    Reto Berra – the Flames tankmachine. It seems kind of silly to me how eager the Flames were to keep trotting him out when (to my eye at least) he was being non-trivially outplayed by Ramo.

    Incidentally, the Reto Berras of the world are the reason I have concerns about the use of PDO as some measure of ‘luck’. There’s quite clearly (to me at least) some element of skill in there.

    Also, nice graphs – what did you make them in?

    • mattyc

      numbers app on osx. generally i think that a pdo range of 985-1015 is “normal” over a season because as you say there are the reto berras and the sid crosbys and tuukka rasks etc.

    • SmellOfVictory

      PDO is primarily a measure of luck, but it is accepted that it is affected to small degrees by skill, both offensively and defensively. For example, the Penguins have a high PDO season-to-season; this is because their top two lines feature Sidney Crosby and Evgeni Malkin (elite offensive centres who boost on-ice shooting percentage) as well as James Neal (an elite shooting talent).

      Additionally, other teams may have high or low PDOs season-to-season because of goaltending (Boston, for example, has had league-leading goaltending for multiple seasons, and thus gets a boost to the ol’ PDO on the defensive side of things).

      That aside, there is still a tendency for PDO to hover at around 1000; team abilities in goaltending and shooting may boost or drop it by up to ~1% (i.e. 999-1010), but it’s rare for the effect to be any greater in the long term. And normally it’s a goalie either sinking or floating the PDO (Kiprusoff was always great at either bolstering the crap out of, or completely destryoing, the Flames’ PDO, for example).

      • mattyc

        Right, each team should regress to their PDO mean (which should be close to, but not exactly, 100%). My concern, is that the value of any stat is to have a certain degree of predictive power. PDOs predictive power is weakened because of Sidney Crosby and Reto Berra. Since Crosby and Berra have a measurable effect on PDO, and PDO correlates well in the short-term with winning, it follows that Crosby and Berra have a disproportionate impact on the result of a game.

        I’m willing to concede that most players should fall within the 100% PDO range (the percentage driven impact of say Stempniak and Moss is negligible), but I’d wager the game is won or lost in the long run, on the back of Crosby or Berra.