44Glen Gulutzan
Photo Credit: Brace Hemmelgarn / USA Today Sports

The Calgary Flames’ numbers don’t match their narrative – and maybe that’s okay

There are a few films that really cut to the heart of what sports is all about. For me, there are two that really get it and both are baseball films: Bull Durham and Moneyball. Both come at baseball from completely different directions and the perspectives of both can provide a bit of insight into just what’s happening with the Calgary Flames lately. Specifically, these two films can provide some perspective in terms of narratives, their limitations, and when those limitations don’t seem to matter.

On trade deadline day, Flames general manager Brad Treliving had his regular interview segment on Sportsnet 960 The Fan. During that segment, Treliving had a line that tipped his hand when he was speaking about the improved play of his goaltenders: “When I really analyze the team, and say if we can strengthen the group in front of them, deepen our defense and give up less chances, it’s amazing how your goaltending becomes a little bit better.”

Treliving is a savvy guy. He gets advanced stats. He’s probably read Moneyball and at the very least saw the trailer for the movie. He might not cite specific numbers or concepts like head coach Glen Gulutzan does from time to time – regularly talking about uncontested zone entries, zone starts or other such things – but he has a very good handle on analytics. Based on Treliving’s assessment, you’d expect that the moves that the Flames made prior to the trade deadline – bringing in Michael Stone and Matt Bartkowski and swapping out Dennis Wideman (to the press box) and Jyrki Jokipakka (to Ottawa) – would result in improved defensive underlying numbers for the team. You’d be wrong, though.

Since arriving in Calgary, Bartkowski has been worse than Jokipakka defensively by every possession metric. Stone is worse than Wideman defensively by every metric except for Scoring Chances and High Danger Scoring Chances Against (Per 60). The Flames replaced their deficient players with guys who are technically downgrades. Team-wide, looking at the past seven games with the current lineup configuration and comparing them with the seven games prior to Bartkowski’s arrival, a similar pattern emerges.

  • Corsi For Against Per 60: 55.56 (now) vs. 54.09 (then), an increase
  • Fenwick Against Per 60: 40.07 (now) vs. 39.40 (then), an increase
  • Shots Against Per 60: 29.91 (now) vs. 29.47 (then), an increase
  • Scoring Chances Against Per 60: 23.84 (now) vs. 25.36 (then), a decrease
  • High Danger Scoring Chances Per 60: 8.53 (now) vs. 10.60 (then), a decrease

The “replacing the defensemen fixed the Flames’ defensive game” narrative doesn’t really pass the smell test.

There’s also been a lot of talk around the rink about the Flames’ turnaround not beginning when the defensemen were swapped out, but back when Gulutzan tore into the team following their “pathetic” performance against Montreal on Jan. 24. The Flames must be markedly better over the last 15 games than they were in their previous 51 based on their torrid 12-2-1 record since then. Again, the numbers don’t really support that argument. Comparing their most recent 15 games with the 15 games prior, the Flames’ offensive underlying numbers are slightly better (an increase of two scoring chances per game) but their defensive underlyings are all flat or worse than they were before. The only thing that was significantly improved was their PDO, which jumped from .949 to 1.031 as the puck stopped going into their net and started pouring into the opposition’s. Once again, the narrative doesn’t pass the smell test.

Maybe it’s about the team’s players finally snapping into place and slotting into the roles many have expected them to play? Particularly on the right side, guys like Micheal Ferland, Troy Brouwer and Alex Chiasson seem to have finally settled into places in the lineup that many of us expected them to based on their performances earlier this season. The general manager mentioned this possibility at the trade deadline.

“This is starting to look like the team that we had hoped it would look like several months ago,” said Treliving during his appearance on Sportsnet 960 The Fan. “Now it’s taken some time, but where people have fallen into roles, it’s starting to look like some of the projection to have in terms of where people slot.”

After the Flames’ 5-2 win over the Islanders on Sunday, Gulutzan had an assessment of his team’s recent performance and how the club’s been all-hands-on-deck lately: “We use guys, we give guys roles, penalty killing, power play, five on five assignments, and maybe put guys in some awkward positions at times and hoping that they succeed. You’re trying to build up your group at all points. We just didn’t want to be a one-line team. You need 20 guys to win, you really do.”

If you’ve been following our post-game recaps or post-game embers, you’ve probably noticed that the stats tables look awfully similar to how they’ve looked much of the season. The Flames are still primarily seeing their possession driven by the same players that have been doing it all season, though their lesser lights are getting lit up a bit less than they were before. If players are snapping into their proper roles, it’s not reflecting itself in the underlying numbers on a game-to-game basis.

That said, if you watch the team on the ice over the last while there seems to be a difference. They don’t collapse when they give up an early goal or a bad goal. They don’t wilt with frustration when they take a penalty. They play largely the same style of game in all three periods of most games, rather than altering course depending on what the scoreboard says. Off the ice, the group seems extremely loose – something made most evident by the recent string of Dougie Hamilton photobombs during post-game television interviews.

For whatever reason, the Flames are winning games and feeling confident right now. It isn’t like the 2014-15 season, either; steps in a positive direction have been evident, even if they aren’t fully there yet. Gulutzan has thrown out terms like “belief”, “commitment” and “playing the game the right way.” They’re playing some of their best hockey of the season right now, regardless of whether that’s being reflected in their underlying numbers. Fans of Moneyball will be naturally suspicious of what, precisely, is happening and why (and it’s a pretty natural reaction). But that approach may risk missing the forest for the trees. For those who’d rather not get bogged down in the numbers, Bull Durham‘s Crash Davis offers a more stout assessment of the roll the Flames are on:

“A player on a streak has to respect the streak… If you believe you’re playing well because you’re getting laid, or because you’re not getting laid, or because you wear women’s underwear, then you are.”

The Flames are playing well. They’re winning games. They’re probably going to make the playoffs. The numbers don’t support any of the narratives about why these things are happening. Does it really matter right now?

  • cjc

    I think the whole point about looking at shot-based metrics is their relevance over large sample sizes. If you look at Calgary’s team possession (a little above 50%), individual possession numbers, and goalie save percentages, you’d probably peg their record about what it is now, give or take a couple points. The only narratives you can make about short streaks are those that don’t rely on numbers, because you can’t separate the signal from the noise without a large enough sample size.

    The season-long narrative is what matters though, not the last 10 games or whatever, and that’s when the underlying numbers start to shine.

    • reidja

      Totally agree with this. However, that PDO swing is huuuge and without looking into SV% vs SH% over the season, I don’t think it’s a stretch to say that the improvement in goaltending is largely responsible for the Flames possession results and place in the standings coming into alignment.

  • Parallex

    Not really (about whether the narrative matters). Really this is mostly just the Luck Dragon showing what a fickle mistress she is… we got the cold shoulder in the first half of the year and now we’re being bedded twice daily.

    • Avalain

      Perhaps yeah. But you have to take out the first 20 games when talking about luck. The Flames were legitimately bad while they started to learn the new system. I’m very interested in seeing how well they do with a whole season tickets work with (though obviously it’s too early to worry about that. We have a date with the playoffs hopefully waiting for us!)

    • Derzie

      The blog should treat ‘luck’ like any other 4 letter word and block the post. Why not use ‘magic’, ‘voodoo’, ‘hocus pocus’? Using luck as an explanation is blind faith. Luck is an element of most everything but it is not a valid conclusion when there is clearly many unexplained factors yet to explain. Confidence for example. Team resolve. Good health, mental & physical. Can’t easily be quantified but are as real as can be, and surely not ‘luck’.

  • Tanner

    This mostly makes sense to me. I have noticed that the underlying statistics haven’t really changed all that much during this run. The Flames’ FF% is 50.7. Their CF% is 50.9%. Their SF% is 51.4%, and this is where those statistics have been for quite a long stretch of this season. That is really encouraging to me as a fan, especially because the PDO hasn’t gone sky high at just over 100. This suggests that the current results are more what should be expected for the Flames (if the Flames are about average with goaltending and shooting ability).

    Also, I don’t think Michael Stone should be seen as a downgrade on Dennis Wideman just yet. Couldn’t it be possible that he is playing a style that allows more shots precisely because it puts emphasis on stopping people from gaining SC and HDSC? I don’t know if that makes sense and it could be that I just have rose-colored glasses on, but perhaps it could be true?

    • Jumping Jack Flash

      Perhaps Michael Stone’s numbers reflect a downgrade from Wideman… But to the naked eye that is not apparent. Of equal importance, to Brodie that does not seem to be the case. As for Bartkowski, this does not seem to be the case. He does not look like he can play at this level. He does not look better than Kulak or W spoon.

      The players are feeling the swagger and that’s all that is important. I really don’t think the fans read too much into the casuals when things are going well. The winning streak could be solely attributed to the H bombs…. It doesn’t matter until it does.

  • T&A4Flames

    A great argument for why analytics need to be taken with a grain of salt. They aren’t the be-all, end-all of why teams achieve success. They look good on the ice. They’re winning and that’s all that matters.

    • BeerCorsi

      No. It’s the exact reason why you need analytics. Teams can go on runs like these all the time and it be beyond stupid to just believe that everything lasts forever and everything is fine just because you’re winning. Had the Flames followed this logic of yours, when they were losing earlier in the season they would have blown it up. But the underlying numbers were saying something else and low and behold they got some luck. The worst style of management a team could possibly have would be one that is not critical at all because the team is on a winning streak.

  • al rain

    A life lived by Bull Durham Wisdom would be a life well lived.

    Interestingly, the other part of that quote, the part in the middle is: “You know why? Because they don’t happen very often.”

    Like Dougie, I’m enjoying the ride. And looking forward to puck drop against the Habs.

  • Jessemadnote

    “Stone is worse than Wideman defensively by every metric except for Scoring Chances and High Danger Scoring Chances Against (Per 60).”

    I would love to see you guys explore this concept a little more. I believe when it comes to shutdown defensemen, corsi against is a very poor evaluater of their play.

    My theory, (Though I don’t have the skill to test it) Is that a good way to evaluate shutdown Dmen would be scoring chances and high danger scoring chances as a fraction of corsi against.

    Case and point: Subban has much better shot suppression numbers than Weber but most people will subjectively tell you that Weber is better defensively.

    Think about it:

    Patrick Kane comes down the wing, Subban pinches, Kane roasts him, goes in uncontested and scores: one corsi event against for Subban.

    Patrick Kane goes down the wing, Weber blocks off his passing lane and forces him to the outside, Kane attempts a weak shot from a bad angle, collects the puck tries another bad angle shot passes back to the point, another shot from far out, Weber is in the lane to block it, settle it down and passes out of the zone back up the ice. Three corsi events against.

    Which player was better defensively in that example?

    • Temple

      Agreed – quality of shots matters, yet isn’t captured by corsi. Also, High Danger Scoring Chances is more subjective than Corsi, but more readily observed (i.e., easy to see a defenceman lose his man on quality opportunities).

      Stone isn’t that much better than Wideman, maybe, but he makes less heart-stopping, bonehead plays. And as you show above – is better at decreasing scoring chances and high danger scoring chances.

      And honestly, as a team, they’re averaging 1/2 a shot more against per game. Gasp! That’ll eventually translate into a goal. Or they’ll suppress the shot total of the Canadiens, and it’ll go back down.

      Don’t read too deep into the tea leaves. Go Flames!

    • Nick24

      The notion of “shutdown defenseman” seems to be where people really get caught in the weeds, however, to be fair, calling the roll “defensemen” is also kind of loaded in it’s own right. Players on the back end should be thought of more as “backs” as although they are there to stop the opposition, a bigger part of their role is the ability to transition the puck up the ice and to then generate chances on the opposition.

      Now, in the case of your example, sure. Maybe Subban makes an ill advised pinch, and the puck ends up in his net. But for every one of those cases, there will be situations when Weber is too slow to catch an attacking winger, gets blown by, and the opposition scores.

      Now if we look at actual goals for and against rates, Weber’s are better. Weber has a 57.14 GF% at 5v5 vs Subban’s 51.14 GF%. However, if we look a little deeper, it becomes clear that Subban is doing more for his team. Subban controls scoring chances at an incredibly high rate of over 58%, where as Weber is at 53.7%. There is value in a player that can more quickly recoup the puck from the defensive end and move it up the ice, and that is something that the prototypical defensive defenseman are less capable of. Shot quality does matter, but so does shot quantity. Over the course of a game, giving up slews of poor chances is just as bad as giving up a few high danger chances. Regardless of quantity, creating chances in the oppositions end is better than the opposition creating chances in yours. That is where players like Subban are so valuable. Keeping the puck in the opposition’s zone is preferable to playing good defense in your zone, so having players who can transition the puck is better than players who are simply “good in their end”. Yes, maybe Weber limits the quality of chances against, but he doesn’t add to the offence in the same way that Subban can, and does. When evaluating defensemen, it’s critically important to look at how well a player can move the puck out of their end and get it up the ice. In today’s game, teams who play in their end are the ones who get scored on.

  • class1div1

    I am one of those nerds that works at processing signals.These contrasts are analog and digital in my crazy world.I compare each of us to an analog waveform that is unique and complete with frequency that is superior in some ranges and weak {noise}in others.Along comes digital {analysts} and takes samples of that waveform ,turns it into a binary code and removes a lot of the noise. Bigger sample sizes improves the waveform even more.This digital tool allows us to focus on weaknesses and improve them, although it doesn’t eliminate the original noise and never will. Personally I love the original analog signal with all its noise. I cant stand digital music .Just a bunch of zeros and ones that could never be duplicated by a musician live.

  • icedawg_42

    I trust that the advanced stats aren’t lying in that there may be no ‘marked improvement’ in this specific area, but come on. There’s no one on earth that’s going to convince me that Wideman is better than…anyone (Save Niklas Grossmann). All you have to do is watch him skate, seriously – he gets beat over and over and over again. It doesn’t take any statistical metric to SEE plainly that Stone is an improvement.

  • Greatsave

    Like it or not, wins and losses are still mainly correlated with, if not caused by, PDO. Does that make advanced stats or possession hockey irrelevant or less relevant? No. I’ve said before that the Flames’ PDO was due for a correction, as over 82 games it is unlikely for a team to finish with a sub-97 PDO (I believe they were sub-97 in 5v5 tied situations, maybe 5v5 overall). This streak is a function of that PDO correcting or regressing to mean.

    Regarding Stone vs. Wideman, it’s important to note also that Stone has seen tougher circumstances than Wideman. Wideman’s 5v5 OZ-to-DZ starts ratio is 1:1, whereas Stone is 0.8:1.

    One concerning trend for me on this streak is the overall 5v5 OZ-to-DZ starts ratio of the whole team. Looking strictly at the past 7 games, when they’ve had three steady D-pairings (5-27, 7-26, 44-29), the ratio has been under 0.75:1. Meaning for every four DZ starts, the team only get three OZ starts. Now, that might be a function of the team leading for a higher proportion of game time over this streak, as leads generally result in a conservative style which leads to fewer OZ starts, but I don’t have the game state times (lead/tie/trail) in front of me so I can only speculate.

  • The Doctor

    I’m not sure if it was mentioned yet, but isn’t part of the recent success an example of reversion to the mean? Our team shooting percentage was inordinately bad/low for the first half of the year, and I understand that it has improved recently. I certainly noticed during the first 40 games or so, we had a noticeable inability to finish — we were getting possession, but in the offensive zone, a lot of things fizzled — passes flubbed, pucks turned over, shots going wide or shots that the opposing goalie could easily see or stop. The other thing related to the improvement in converting scoring chances is the fact that these guys have now had some more time to play with one another. The Flames have done, fundamentally, what you are supposed to do over an NHL season: gradually work your kinks and bugs out, get settled and get more confident and proficient.

  • dontcryWOLF88

    How come “post comment” is at the top now. Was better at the bottom where the conversation is going on.

    Speaking of that conversation. And the article.

    There is no stat line or point of comparison for measuring confidence. Since this is arguably one of the most important aspects of any form of human achievement, that’s a big hole in input data.

    Example? Anyone ever play a game of gold or pool and be lighting it up. Then have one bad shot and watch it all fly out the window? Since confidence is not purely random it is the factor confounding statistical of anything human related.

    • The Doctor

      “There is no stat line or point of comparison for measuring confidence.”
      Exactly. I’m reminded of Ken Dryden’s book The Game, which was written in the context of his being part of one of the greatest NHL teams ever (and I say that as a non-Habs fan BTW). In the book, Dryden talks a lot about that, how those mid to late 70s Habs teams, no matter what was going during a particular game, always had the confidence that they could turn things around.

  • Derzie

    In the pursuit of Wins, goals are must-have, scoring chances are should-have, Corsi is nice-to-have. On both sides of the puck (offense and defense). Removing Wideman and moving Chiasson away from the top lines is having an immediate impact on Wins, Goals & Scoring chances. Because it is a ‘streak’ it is not sustainable but the moves clearly had the desired impact in the short term. We’ll talk Corsi after the dust settles (next season). But only if the more important indicators are trending nicely.

  • beloch

    Stone is worse than Wideman defensively by every metric except for Scoring Chances and High Danger Scoring Chances Against (Per 60).

    To be fair to Stone, those are two key defensive metrics. Improve both of those and it’s expected that your goalie’s save percentages will climb.

  • Jumping Jack Flash

    I think it is safe to say that if the Flames had to call up a keeper from Stockton, Ritiich is the man right now. Another shutout last night giving him 5 for the season has put him in a position to be the go to guy down the stretch. I expect Huska will ride Ritiich for a series of games. This will be good for Gillies who is playing some of his best hockey this year as well.

    Just more good news for the organization.

  • “Now it’s taken some time, but where people have fallen into roles, it’s starting to look like some of the projection to have in terms of where people slot.” — Treliving. I disagree on this one point. I don’t think Treliving means that this is where they expected each guy to fall in to the line up when they drew up their lines at the start of the year. If that’s the case then we have a bigger problem on our hands both with the disjointed communication between management and coaches and with the managements perception of what to pay for 3rd and 4th line right wingers.

    If Treliving had envisioned Brouwer playing on the 4th line when He signed him to that 4.5 million dollar contract then Treliving far too heavily over values experience and leadership and right handedness as traits for hockey players. Treliving doesn’t seem like this kinda guy so I’m just gonna go ahead and call nonsense on that. I suppose you can’t expect a guy to come out and flat up say hey we made a huge mistake on paying the 30 year old with stone hands a 4 year 4.5 million deal. OH well.

    • Jumping Jack Flash

      I still feel that the real Brouwer will show up when it matters. I also agree that 4.5 million a year is too much to pay for his output. People talk about confidence all the time but it is not just a barrier for young players breaking into the league….veterans like Elliott and Brouwer have looked lost at times.

      Brouwer might have to pull a Maroon over the summer and lose some mass so that he can gain some quickness.

  • Lucky 13

    I really enjoyed watching MoneyBall and the metrics used by the Oakland A’s for putting together a low cost team that won 19 games in a row!

    Too bad they didn’t win the World Series, however Boston used the very same metrics and won a World Series two years later, albeit with a much better cast of players.

    But I’m a Canadian and I love our Jays, so I’m going with Marcus Stroman’s philosophy ” HDMH” height doesn’t measure heart.
    Strategy results from being able to measure or track progress (stats) but you can’t measure success without having the core or heart of the person whom it resides. That’s why it’s important to keep in mind that one or two players don’t make a team, but 23 do.
    That’s why teams like Oakland were successful and defied all logic in 2002. They all believed in each other and that’s sometimes more important than just analytics alone.

    Now we are on a 7 game winning streak. Imagine if we could win 19 in a row!!

    • The Doctor

      Another thing I really liked about Moneyball (the book and the film) was the way that it offered a well-deserved critique of groupthink and conventional wisdom. Michael Lewis and Billy Bean both realized that sometimes, just because everybody thinks something or everybody does something does not make it correct. Specifically Bean and Lewis saw that baseball scouts (and hockey scouts have the same bias) tend to go ga-ga over impressive physical specimens. Guys who are big, fast, graceful, athletic and so on. But often those guys don’t have the secret sauce that makes a great player of the particular sport. And of course the NHL is full of guys who either weren’t drafted at all, or were drafted really low, that just plain knew how to play the game well, even though they weren’t physical studs — Marty St. Louis, Mark Giordano and Johnny Gaudreau being quick examples.