Flames Expectations First Update

At the beginning of the
season I took a look at the Flames and ran some numbers like sh%, ppg pace,
historical games played and the like. Using that information I estimated the
number of games played, estimated sh%, sv%, and so on for the Flames players.

Most times we stop at the
quarter-points of the season to assess a team. I’m choosing to stop at the
one-third marker in order to provide a larger sample size and to try to bridge
some of the borders that exists within a season, to blur the edges so to speak,
so that the data may have a bit more applicability to the larger season rather
than being a snapshot of the here and now.

There is quite a lot of data,
so bear with me, but I hope that what will become clear is the enormously high
percentages that the Flames are currently riding (and their fan base thereby

Here is a link to the
original articles for forwards and defense.

Now, to the Great Tableau of Data (I apologize for the ugly format)


Player est. gp   actual
est. ppg hist. ppg hist.
  to date
 pts pace
Hudler 80 25 78 0.7 0.56 14.2 26.2 56 25 78
Stajan 70 11 35 0.48 0.48 13.3 11.1 33.6 2 7
Glencross  78 26 82 0.6 0.55 15 8.9 47 16 50
Raymond 75 10 31 0.5 0.49 9.7 20.8 38 7 57
Monahan 75 26 82 0.4 0.45 12 13.2 30 17 53
Colborne 80 11 35 0.35 0.35 12 0 28 8 57
Jones 70 13 41 0.4 0.45 13.7 16.7 28 6 20
Backlund 73 11 35 0.5 0.41 8.5 5.6 37 4 12
Byron 70 26 82 0.38 0.36 15 13.9 26 10 26
Baertschi 41 13 41 0.42 0.47 8.7 0 17 3 9
Bouma 73 25 78 0.18 0.16 5.1 20.8 13 9 28
Bollig 70 24 75 0.15 0.11 4.4 0 11 2 6
McGrattan 60 7 22 0.1 0.09 5.1 0 6 0 0
Van Brabant 15 0 0 0.08 NA 0.05 0 2 0 0
Agostino 12 0 24 0.2 NA 8.3 0 2 0 0
Gaudreau 70 25 78 0.55 NA NA 8.9 38 19 59
Wolf 15 0 0 0.05 NA NA 0 1 0 0
Arnold 12 0 4 0.25 NA NA 0 3 0 0
Knight 20 2 7 0.35 NA NA 0 7 0 0
Granlund 20 15 48 0.35 NA NA 13 7 9 29
Jooris 10 19 60 0.25 NA NA 27.6 2.5 12 38
Hanowski 20 0 34 0.25 NA NA 0 5 0 0
Reinhart 12 4 12 0.26 NA NA 0 3 0 0
Ferland 10 7 22 0.29 NA NA 0 3 0 0
Bennett 9 0 0 0.33 NA NA 0 3 0 0
Setoguchi 73 12 37 0.48 0.54 11.4 0 35 0 0
Giordano 82 26 82 0.53 0.44 7.1 10.3 43 26 82
Brodie 82 26 82 0.32 0.46 4.1 13.3 26 21 65
Wideman 75 25 78 0.47 0.47 6 15.1 32 15 46
Engelland 75 21 66 0.14 0.19 5.7 0 11 2 7
Russell 70 24 75 0.35 0.29 5.2 0 25 10 31
Diaz 42 10 31 0.25 0.29 3.6 0 8 0 0
Smid 76 24 75 0.11 0.13 3.6 0 8 1 3
Wotherspoon 30 0 0.15 0.19 (AHL) 0.5 0 5 0 0

Player    est. gp   gp pace   est. shpg   shpg actual   est. sv%   sv% actual  est ga/game   ga/game actual
Hiller 68 50 28.5 28.1 0.912 0.913 2.5 2.49
Ramo 14 32 28.6 27.4 0.905 0.921 2.7 2.25
0.9085 0.917 2.6 2.37

(Player data taken from Flames website)

I had the Flames pegged at
about 206 goals for this season. Their current pace puts them at 253 goals for
on the season.

Hiller’s actual sv% is out by
0.001 and Ramo’s is ahead by 0.016, substantial improvement over his career
average. The goals against per game was estimated at 2.6 on average and the Flames
are currently averaging 2.5 goals against and 3.08 goals for per game. Ramo is
playing more often than anticipated, obviously on account of his strong play
thus far, and is on pace to play 32 games this season.

The rest of the data more or
less speaks for itself. Giordano is on a ppg pace with Brodie not far behind.
Wideman, surprisingly, isn’t actually off his career pace.

As expected, Jiri Hudler
looks like the forward offensive engine of the team with a collection of other
players chipping in offense, albeit at a far more productive level than
anticipated, and there are a few players whose offensive output has been underwhelming
thus far but are being sheltered by the output of others in the roster or are
seeing diminished results due to injuries.

As of December 3rd,
when this information was collected, Calgary had four players in the top 30 in
the NHL for shooting percentage for players who have played more than 10 games.
All of those were above 20%. No other team had that many. Those players are
Jooris, Raymond, Hudler and Bouma, one core player, two complementary talents
and a rookie.

I’m going to include a table
here showing PDO history for seasons dating back to 2009-2010. I’ve selected
the teams from each season whose PDO was above 100 and tracked them the
following season. I’ve marked the 2012-13 season with an asterisk to denote the
lockout-amended season and put an exclamation point in front of this season because
we are only one-third complete.

The Finish column denotes
where the team finished while the Following column marks where that same team
finished the following season. I’ve kept Conference standings rather than
overall standings to highlight the difference in competition between the two
conferences and thereby reflect the perception that the Western Conference is
the more difficult of the two.

In the final column I note
the difference in standings from one season to next with the coloured cells
highlighting positive adjustments from one year to another and all others being
negative adjustments. The majority of the events show a regression in the following
season, with some season-by-season eccentricities such as a universal
regression by all teams in 2012-13 and a 50/50 split on regression in 2013-14.
The sum total of all standings adjustments was -35, meaning that the overall
trend was for teams to lose ground the following season, while the average
standing loss was -1.4 places following a PDO peak with the most common
occurrence being a setback of 1 place in the standings.

Before we begin, I’d like to address PDO as a concept. I think this statistical category is useful, but misunderstood. Like Corsi, Fenwick and others, it is one tool in helping to illustrate a larger picture. When we say that PDO historically regresses towards the mean, it doesn’t mean that every single team will ultimately come back to 100. Some teams are better than others for long stretches and there is always the role of the exceptional individual to account for. But, by and large, teams that have shooting percentages that sit at 15% or a save percentage around 0.941 for a season are unlikely to sustain that peak for too long a time. The league is too good and the parity amongst teams far too strong for that kind of advantage to be maintained.

Also, the mean that we refer isn’t as strict as many believe it to be, just as strong possession numbers are an indicator of a good team but not necessarily a proxy for one. Mikael Backlund and Kyle Brodziak have been possession darlings in the past and neither was ever in danger of being voted to the All-Star team as a result of it. But both are very good players to have on a team, and successful teams usually have one or two of those guys on them, and if they have elite-level talent as well then those teams tend to win championships. It is a correlation, not a causation. With PDO, don’t look for the teams that are always at the top of the list, instead check for the teams making a cameo. The exceptions, in this case, often prove the rule.

Think of it like weather. The average daily temperature for Calgary in December is -2 Celsius. It may never actually be -2 C in Calgary during that month, but on average the temperatures during the day will even out to somewhere within that range. There will be hotter days and colder ones, just as there are better and worse teams in the NHL. The good teams have better players who may have higher shooting percentages or goalies with higher save percentages. It comes back to the basic philosophy of sport: to win you need better players. Going back to our weather analogy, if you woke up one mid-December day and it was 15 degrees above you probably wouldn’t think that Calgary was now in an equatorial climate zone. You are more likely to think that there are some funky weather patterns messing things up and would expect it to go, more or less, back to normal in a little while.


Team  PDO      Sh%    Sv%     Finish       Following   PDO       Difference
2009-10 Wsh 103.15 10.30 0.928 1st EC 1st EC 100.18 -1
Van 101.71 9.35 0.923 3rd WC 1st WC 101.53 2
SJ 101.65 8.71 0.929 1st WC 2nd WC 100.21 -1
Col 101.50 8.89 0.926 8th WC 14th WC 99.33 -6
Ana 100.99 8.19 0.927 11th WC 4th WC 100.13 7
2010-11 Bos 102.43 8.29 0.941 3rd EC 2nd EC 101.41 -1
Nsh 101.62 8.17 0.934 5th WC 4th WC 101.04 -1
Van 101.53 8.19 0.933 1st WC 1st WC 101.35
Phi 101.38 8.74 0.926 2nd EC 5th EC 100.44 -3
Phx 101.32 8.02 0.933 6th WC 3rd WC 101.18 3
Dal 101.19 8.69 0.925 9th WC 10th WC 99.83 -1
2011-12 Det 101.50 8.75 0.927 5th WC 7th WC 100.19 -2
Bos 101.41 9.02 0.923 2nd EC 4th EC 100.51 -2
Van 101.35 8.20 0.931 1st WC 3rd WC 100.59 -2
Phx 101.18 8.03 0.931 3rd WC 10th WC 100.35 -7
NYR 101.08 8.35 0.927 1st EC 6th EC 100.92 -5
Nsh 101.04 8.47 0.925 4th WC 14th WC 99.54 -10
*2012-13 Tor 102.99 10.67 0.923 5th EC 12th EC 101.20 -7
Pit 102.82 9.66 0.931 1st EC 2nd WC 100.06 -1
Chi 102.03 9.08 0.929 1st WC 5th WC 99.87 -4
Ana 101.72 8.65 0.931 2nd WC 1st WC 102.32 1
CBJ 101.63 8.70 0.929 9th WC 7th EC 100.72 2
Wsh 101.41 8.61 0.928 3rd EC 9th EC 100.16 -6
Dal 101.29 9.47 0.918 11th WC 8th WC 100.34 3
TB 101.23 9.80 0.914 14th EC 3rd EC 100.58 9
Mtl 101.03 8.93 0.921 2nd EC 4th EC 100.54 -2
2013-14 Bos 102.46 8.46 0.939 1st EC
Ana 102.32 9.79 0.925 1st WC
Col 101.75 8.80 0.929 2nd WC
Tor 101.20 8.40 0.927 12th EC
Min 101.13 7.91 0.932 7th WC
!2014-15 Pit 102.89 9.17 0.937
Nsh 102.90 8.26 0.946
Cgy 102.37 9.78 0.925
LA 101.80 7.96 0.938
TB 101.41 9.82 0.915
Tor 101.19 9.59 0.916
NYR 101.12 9.21 0.919
StL 101.00 7.17 0.938

(PDO data taken from War on Ice)

What this data means for the
Flames is that they are likely experiencing a PDO peak (aka lightning in a
bottle) and can be reasonably expected to have a regression. Unless you believe the Flames, as a team, are going to maintain a shooting percentage of 9.78 (league average this year is 8.85 and hasn’t been above 9.5% since 2006-07, in fact this site has the Flames’ sh% at an astonishing 11.54%).

The Flames’ save percentage is high, but not abnormally so, in my opinion. I would suggest that Pittsburgh, Nashville and St. Louis are likely to see a correction there from their lofty sv% heights, less likely with L.A. based on the history of Jonathan Quick as a player and the L.A. Kings’ playing style in limiting scoring chances against.

This does not
necessarily mean that a Flames’ regression will occur this season and when it does eventually, because it
will, it may not necessarily be a catastrophic collapse into the bottom of the
league (think Colorado over the past few years), but may result in the
team hovering around a few places mid-standings.

There are quite a few factors
waiting to play out, namely consistency in this season personnel and asset
decisions at the trade deadline and the eventual inclusion of prospects like
Sam Bennett, so we will see how the team evolves over the next few months.

Why this song?

Heavy Fuel isn’t one of my favourite Dire Straits songs. Mark Knopfler is a tremendous talent and On Every Street is a great album, but this track always felt out of place, put there by record executives who needed something radio-ready. That being said, the song fits for the Flames. It is about a working man leading a hard life, earning his pay and enjoying every ounce of good fortune that comes his way, deserved or not. Others warn him of the cost of his hard-driving lifestyle, telling him it is unsustainable, but the lure of the road and the joys that come with it are more rewarding.

There are a lot of voices telling the Flames and their fans that this can’t last and they are probably right, but you can’t tell a Flames fan looking at the standings and dreaming about playoff dates and a C of red in May not to be excited about what this team has accomplished.

  • piscera.infada

    There are a lot of voices telling the Flames and their fans that this can’t last and they are probably right, but you can’t tell a Flames fan looking at the standings and dreaming about playoff dates and a C of red in May not to be excited about what this team has accomplished.

    Thanks Rex. Sums it up perfectly. Interesting that we need an Oilers fan to tell us this.

    Good read. I always appreciate your work, even if you’re one of them…

  • Derzie

    The Flames play a different style than the average NHL team therefore their stats are not to be measured against average. The Flames play should be pushing statisticians to challenge their beliefs rather than hang on to them as if they have talisman-like powers. Flames will have losses to be true but barring injury, this is the team and what they can do. Create new measures to find out why, don’t hang on to established thresholds and means.

  • Derzie

    Last year was billed as Advanced Stats’ “coming out party.” I wonder how they will be viewed this year? Maybe it’s mostly Flames fans, but this seems like “the year of pushback.”

    I don’t disagree with the fancy stats and I think it’s pure hubris on the part of a lot of fans to think that the Flames have suddenly found some new way of playing hockey that is statistically unquantifiable or can somehow ‘beat the stats.’ *apolgies to Derzie*

    That said, while I disagree with one part of Derzie’s post, I tend to agree with the other part. That is, and many others have stated the same, there really seems to be a ‘good enough’ mentality among a lot of the advanced stats crowd. Instead of just subscribing anything outside of the ordinary as “luck,” why not look for the mathematical answer?

    It sometimes seems to me as though the fancy stats crowd is largely made of people who have an ‘above-normal’ interest in math and one day they happened upon Corsi and Fenwick and looked at it and said, “good enough.” And that’s where it stopped. Those stats are right most of the time and any deviation is luck and “goodbye.” “I said goodbye!”

    What needs to happen for advanced stats to truly emerge, is for someone to come in and explain why the Oilers suck despite their possession numbers and why the Flames are so much better than their possession numbers. That would be groundbreaking.

    For example, I think it’s naive to say that the Flames simply ‘outwork’ other teams, as though work ethic only exists for one club in the NHL. However, by the eye test, that work ethic seems to come into play late in games, so are other teams taking their foot off the gas late into games or after leads? If so, will that last? My gut says “no.” It’s almost like the Flames are already playing playoff hockey; what happens as other clubs start to ramp things up themselves? Would quantifying such moments in time even be useful? I don’t know, but the point is that more investigation is required.

    As it stands now, the comment from Burke about, “being more useful for support than for illumination,” is starting to make an awful lot more sense. Therefore, deeper analysis is imperative. Why? Because I’ve seen about a dozen boatloads of Flames fans go from advanced stats adherents to fan boys almost overnight, simply because it’s their team now defying the stats. My big fear with that is what the narratives will become when the Flames do regress. I don’t know when or by how much the Flames will regress, but it will happen at some point. What then?

    For example, look at how many fans were declaring Granlund as superior to Backlund, a multi-season vet. Last year Backlund was golden and now a rookie who had (at the time of the comments) 11 games under his belt was superior. WTH?! An 11 game sample size! Maybe playing with Hudler and Gaudreau helped?

    So, I fear the narratives more than anything. Lambert is right about one thing, the Flames have to stay the course. Whether they tail off this year or implode next year or the players individually improve ala Monahan and the stats start justifying the results, whatever the scenario, DO NOT DEVIATE FROM THE REBUILD, because there will be a great many fans coming up with all sorts of nonsense to explain the situation and it’s then that the advanced stats become so necessary.

    All that said, I will cheer for the team’s success because it is fun to watch and refreshing to win, because I enjoy seeing the youth develop and a winning culture is superior to a losing one. That doesn’t mean I’m ready to cast off sound math and good sense though (because even the eye test shows the Flames are playing over their heads). However, until the advanced stats gurus can come back with some new insights and new analysis to start explaining the outliers, I’ve pretty much had my fill of the “it’s all luck” articles. Give me something more, please.

    • SoCalFlamesFan

      Excellent post. I find it interesting that in an “enlightened” society the first response of statisticians is to call upon “luck” or “chance”, basically calling upon some mysticism, aka hockey gods like a polytheist.

      Why can’t a statistician just admit there are just factors not accounted for yet and that the stats therefore are just “general guides”.

    • playastation

      Agree on all fronts.

      The ability to actually explain the discrepancy between the Oilers and Flames is a huge deal. Let alone the fact that the Oilers currently have a much better Corsi rating than the Kings (most of the league does). Saying, ‘well the Flames are luckier’ is hard for many to swallow. As we’ve all watched Oilers games and watched a bunch of people skate up, enter the blueline and put a shot on net.

      That being said, shooting percentage and shots should be separated into scoring areas, and corsi should be the ratio of shots from scoring areas * percentage.

      For example, 3 scoring areas = Blueline, Slot, Crease and
      3 Shooting Percentages = Blueline, Slot, Crease

      Example Flames vs Denver Broncos
      Flames direct 28 shots
      Broncos direct 34 shots

      Lets say the Flames won 1 – 0

      Rel Corsi is -6. Flames must have been lucky.

      Or with something like the weighted average shots * sh% / weighted average shots against * sh%

      Lets say the breakdown of LEAGUE sh is like this:
      BL = 2%
      SL = 3%
      RB = 3%

      And the Flames did something like
      BL = 15 Shots * 2% +
      SL = 8 Shots * 3% +
      RB = 8 Shots * 3% +
      = .78
      MINUS the Broncos
      BL = 30 Shots * 2% +
      SL = 2 Shots * 3% +
      RB = 2 Shots * 3% +
      = .72

      Diff = .6

      This would be using something like league wide shooting percentages. Then the expectation here would be that the flames would win .78 to .72 despite losing the corsi battle. This is just an example (its also how we figure out cost of capital in finance) of things that should be explored. We could add in the teams personal shooting percentage (or even the another weighted average of the personal players lifetime percentages in each area) to further come up with how much these percentages deviate from the league wide percentages.

      Let’s remember, regressing towards the mean doesn’t actually mean getting to the mean, league AVERAGE means being within 1 deviation of the mean, which is a range. So some teams sporting a percentage point above the mean isn’t a big deal.

      There are also explanations; the Broncos were mainly kept to the outside, while most of the flames shots came from the slot or below. Regardless of team sh%, analysis can be done. I’m also guessing this is how statisticians on teams would break it down, lets remember, some of the statisticians are actual math guys, and not journalists, but breaking it down and explaining the underlying numbers rather than dividing shots and saying ‘underlying numbers’ is actually less complicated for people to deal with.

      Until things like this are done, I have a hard time taking Corsi or PDO seriously. Especially considering that article that come out about one of the 80’s Oilers team being bad because of their corsi.

      Why. Why are they bad? Because corsi? Come on.

    • RexLibris

      On the furthering of advanced stats:

      We’ve seen occasions of this within the blogosphere where people like Tyler Dellow and others will drill down into a specific topic such as Zone Entries or possession recovery strategies following a DZ faceoff loss and in many cases the individuals who do the work end up becoming employees of NHL teams.

      The data is out there, and the work has been done, but it seems to often become proprietary before it has time to fully mature within an open-source environment.

      Wrt math people, Kent has stated his amateur ranking when it comes to mathematics in his background (I think it was something to the effect of “I wasn’t ever really good at math, but my interest in hockey made me work at it”) and I am very similar.

      Anyone who knows me would be very surprised to see me labouring in front of a spreadsheet parsing data six ways from Sunday. I may be analytical, but I’m not one who has a long history of familiarity with math.

      That being said, I believe strongly that mathematical ability is relatively innate (it is basically logic, after all) and that the difficulty lies not in the subject but in the method by which the student can best grasp it.

      It isn’t that the destination is out of reach, its that the road to get there isn’t always easy.

      The Flames work very hard and to some extent create their own luck. However, when hard work and average talent meets hard work and superior talent, the former almost always loses to the latter. That is often the story of teams that overachieve once they reach the playoffs.

      Those who said Granlund was superior to Backlund may have discussed advanced stats, but that doesn’t make them innately prone to a long-term vision of the team. There are reactionary stat-heads, too.

      Great points, all.

      If nothing else, perhaps this season will give fans an opportunity to sit back and think critically about advanced statistics, narratives, and what these two, seemingly (but not necessarily) contradictory paths mean for fans.

    • Kevin R

      Great post Wolf. Couldn’t have said it better. The only thing is that I think the eye test doesn’t tell me these guys are playing over their heads, so I am not sure what you meant there. My eye test of the games I have gone to & watched is that this team is very fast. When they finally get possession of the puck in our end, they are usually pretty effective in not only getting it out but also going in fast on the opponents. We were discussing it Saturday night that the speed the Flames can counter attack with are putting the opponents back on their heels more. Sharks were trying to trap & that wasn’t working & several times they were getting caught if they were trying to forecheck too aggressively. Flames speed & counter attacking was the reason for this.

      So what seems to be happening is that the chances/shots Flames are getting, are at way better locations & getting prime 2nd chances off rebounds or over committing defenders. This is part of the reason the shooting %’s are up. The other biggest thing is we haven’t seen a player wearing a Flames jersey capable of setting up in the offensive zone like Gaudreau does since the Magic Man Kent Nilsson. Huselius did to some extent but nothing like Magic Man & Gaudreau has so far.

      If a bunch of our wins were like that Chicago game at the beginning of the year, I would have picked up a stick & started beating that regression drum myself. Something else is at play here & you are right, the stats crowd need to take this to the next level. Personally, how fast of a team the Flames have become has just really hit me this year.

      Agree, at no point do we become buyers, playoffs or not. Stay the course & what we don’t get in a higher draft pick we parlay the increasing value on our assets during this success to important pieces for this team moving forward. I love the weather analogy Rex, but the stats people need to evolve & adapt to a larger event that may be unfolding like global warming/climate change. I have lived in Calgary all my life & truly when I was young April showers would bring May flowers. Not any more & hasn’t for the last 15 years. The mean now has changed to June showers(floods) bringing July flowers. Enjoyed the read!

    • Captain Ron

      It’s comments like this from you and some others around here that have kept me coming back to this site for some time now. I have come to realize that it happens quite often where I enjoy the comments section more than the article itself. This is a fine example of why.

    • mk

      If you’re looking for people who are trying to quantify the reason some teams outperform or underperform their possession stats, there are many attempts at analyzing shot quality.

      Unfortunately, it is VERY difficult to do well: much for the shot location/shot type data is inaccurate (NEVER use NHL shot location data -> their scorers can’t even make down the location of the face-off dots correctly, and its painted on the ice). So the only people that can do it well are those that spend the time to manually find their own data OR are using some comp-controlled video analysis.

      Some people will also say shot quality isn’t sustainable because of how they look at shot quality. An example: there is a correlation between shot distance & quality (closer is better), but it is strongly suggested by the data that shot distance is not a consistent ability (i.e. players will not consistently be able to shoot from a closer distance). The conclusion is that shot quality is only very rarely a skill (only a few players can defy the averages by a lot for a long time).

      However, other definitions of shot quality show that shot quality can be system dependent, especially when looking at the shots a ‘tender faces behind a specific team (such as looking at the circumstances of the shot: clear view of the shot, screened, cross-ice pass before the shot, square to the shooter or not, tip-in, etc.). This explains very well why the system the Bruins employ reduces the shot quality of the shots their goalies face -> sustainably driving their PDO high. The fact that they are also strong in possession stats is the reason they are a consistent contender.

    • Parallex

      I think it is mostly Flames fans… ‘pushback’ has always come from the fans of the team that was defying the data. I’ve yet to hear anything from any of the ‘pushbackers’ (for lack of a better term) that I haven’t heard elsewhere from fans of the Maple Leafs and Avalanche. I get a feeling of deja vu back to the 09-10 Avalanche more then anything else.

      On a side note I wish the Fancy Stats had a different naming convention. It’s harder to explain them when the name isn’t descriptive… baseball is way better in that regard.

  • xeno

    I was thinking playoffs 3 years from now but I’d be happy if they gave us a 2015 sneak peak. I can’t wait for this team to be a perennial playoff team!

    For now I’m just enjoying each and every game, you never know what could happen!

    Go Flames Go!

  • RedMan

    as a fan, I am not even getting excited about a possibility of playoffs… not even in the equation.

    What I am pleasantly surprised about though is:

    1) the teams ability to win so many games

    2) development of key players like monohan, gaudrea, ferland, brodie, piorier, etc

    3) the commitment to systems, playing hard every shift, earned ice-time, never giving up and coasting to a finish

    The fact that flames legitimately have people in the talk for the norris, heart, jack adams and rookie of the year… I mean, will any of these win these awards? highly unlikely based on the fact that flames more than likely not in the playoffs plus the eastern bias, but they are ~~legitimately~~ in the conversation.

    Last year we lost a lot of 1 goal games, thanks in a large part to the experimenting done with the three different goalies… McDonald!?!?! we had said then that if the goalies were even even league average over the season, we would have won more of those games.

    Flames have put together a very good start to the rebuild, and there is no denying that!

    some are actually mocking the flames, saying “haha, you think your so good but will crash back to earth, be the next colorado or toronto” Well, this might be a good laugh on us IF we went around proclaiming the rebuild over and our team to have achieved greatness, but in case all the pundits haven’t noticed, NOBODY here or anywhere that I have heard or seen are doing that – we are just enjoying “one day at a time”


  • SoCalFlamesFan

    Good article Rex,

    It’s nice getting a little more stats than just PDO and Corsi, as I feel they don’t quite paint the whole picture..

    The Flames are 9th in the NHL in 5 on 5 Goals for and Against ratio, as well as power play %.

    They are not taking a lot of shots, but not allowing a lot of shots either.

    I’m starting to second guess this “overachieving” and “regression” mindset.

  • RedMan

    If you weight one side of the dice or one side of a coin – you skew the odds.

    If, after weighting the dice or coin, you stick to the original mathematical equation, i.e. a coin toss = 50/50, the PREDICTIVE value of the 50/50 becomes far less accurate.

    if the coin rolls 60/40 to the weighted side and you still stick tot the original 50/50 prediction without factoring the weighted affect, you might deduce that the 60% was “lucky bounces”

    somehow there is a reason the Flames have been “lucky”. Luck has very little to do with it.

  • Purple Hazze

    So looking at your table above between 2009 to the end of the 2014 season:

    -46% of the teams that had a high PDO in season 1 got better or stayed the same the following season

    -If we include teams that only fell one position in the standings from one season to the next my previous number jumps up to 73%

    -only 7 times (30%) where a team had a high PDO in season 1 was there a significant drop in the standings the next season for that team. In 3 of those occasions the team still made the playoffs the following season despite the fall in standings.

    I think these numbers work against the logic that high PDO is necessarily a bad thing.

    • RedMan

      Yeah, that’s what I got from the table as well. Since PDO and corsi have been shoved down our throats the last few weeks, those comparisons are the only fresh material that stood out. Interesting, and not much of a “warning sign” going forward.

    • RexLibris

      The teams that had high PDO for several years running generally had talented teams (Ovechkin, Crosby, Kane, Toews, etc).

      The teams that made cameos sometimes did as well (Duchene, O’Reilly).

      If your band is making an appearance on the PDO top 40 with Kasey Kasem, and all you’ve got are backup singers and a great rhythm guitar, then something is amiss.

      High PDO isn’t necessarily a bad thing, but misinterpreting the results as “we are a Stanley Cup caliber team” when every other factor suggests otherwise (Corsi, Fenwick, salary cap, players on the bench, past history, etc) is a bad thing.

      It is important to know when you are getting lucky and not do anything rash as a result.

      Example: guy goes to Vegas.

      Hits the tables and takes $400 to $40000 in two hours.

      He can a.) think he’s figured out a system that beats the house or b.) take his winnings and leave.

      What does he do?

      a.) doesn’t end well.

      b.) buys his wife a diamond, gets the kids’ college fund topped up, and tells one beauty of a story to his buddies.

      PDO is the cold clear light of day that teams would be well-advised to note, even the ones that have good talent and are performing well.

        • RexLibris

          Ha, I think “cash out” is an excellent metaphor for what some teams will be doing in the coming weeks re: Connor McDavid.

          Wrt the casino metaphor, PDO suggests that all teams over time revert towards the mean.

          Over time, and towards.

          Not immediately and not exactly to the mean, just that the pool of talent within the league and the movement of it within said league means that teams go through cycles and there is a great deal of parity.

          So, if a team finds itself beating the odds significantly, how does that affect their decision-making? Do they think they’ve become better than the rest of the league or do they dissect the facts and try to figure out why they are having so much success?

          The 2008-2009 Oilers (or more specifically the ownership group at the time) convinced themselves that Tom Gilbert, Sam Gagner, Ladislav Smid, Andrew Cogliano and Robert Nilsson were a core group around which they were going to build a Stanley Cup contender.

          Rather than review the season objectively they saw the results and let that convince them the team could be a contender. Enter Sheldon Souray, Tomas Vanek offer sheets and the like.

          They had a PDO of 101.33 backed by a sv% of .925. Most anyone could have looked at that and said that they were due for a correction the following season.

          2009-2010 was that correction. With fries on the side.

  • You do a great job putting PDO into perspective. I still think that comparing team to team, or season to season using these types of numbers is a bit A to O, but still potentially the best thing we have at this point.

    By the way, has anything like this been done before? –
    The error could easily be estimated by going back 3 to 5 years and doing a mock forecast of the future using the advanced stat metrics. Then you could compare the mock forecast to the actual results. This would make Corsi and friends a lot more convincing… well, assuming the error is found to be small. I might give something like this a try once my final exams are done.

    Edit: And the point of this would be: if the error is found to be less than 10%, then you can say with confidence that there is some significant difference between 45% and 55% Corsi. If not, well then we’re back to the same debate.

      • See, that’s still just the tip of the iceberg as far as I’m concerned. It still only takes into account averages… no standard deviations and no error. Comparatively speaking, “Corsi Tied” might be the best indicator of future success. That still doesn’t mean that it’s a good indicator of success… it just means it’s the best one we have.

        I want to see a comparison of Corsi to itself. In other words, predict a team’s success using Corsi, but start with data from 3 to 5 years ago. Then compare the predicted success to the team’s actual success (as recorded in history). This will paint a very meaningful picture.

    • RexLibris

      If you are interested, go to War on Ice and look back at old FenClose leaders.

      Then compare them to the teams that made it furthest into the playoffs for that year.

      I’m not entirely certain off the top of my head, but the correlation is strong, in the 70% range, I think.

      FenClose is the best predictor of success.

      The LA Kings have been excellent FenClose performers the last few years.

      It revolves around Darryl Sutter’s comment regarding possession that, paraphrased, is thus: if you have the puck, then the other guy can’t score. And when you shoot the puck the idea is to either score or get it back again right away.

      That is the crux of all possession metrics right there.

      One puck. Every second you have it is one more second the other team can’t score. Shoot it and then get it back again right away.

      This is age-old wisdom amongst hockey coaches, all things like Corsi and Fenwick do is formalize it and strip away the “he looks like a hockey player” crap that muddies the water.

        • RexLibris

          Because Corsi is a useful tool for in-game and player evaluation. Fenwick is more of a team-evaluation tool and therefore is better suited to determining which team will perform better over a longer period of time.

          Plus the media jocks have just figured Corsi out.

          Start talking about Fenwick and FenClose and you’ll see their eyes glaze over.

          No one number is the be-all and end-all.

          Take everything into account and then make your judgement.

          Imagine you are betting on a boxing match.

          You don’t just pick the biggest gorilla in the ring.

          You look at athleticism, stamina, power, location, recent injury history, past match results versus similar opponents.

      • I definitely need to look into it more. It’s really interesting stuff. I will bookmark this and dig into it over the Christmas holidays.

        Until then, I’ll consider the case closed… but you will definitely hear from me if I find something contrary ;]

        • RexLibris

          I’d recommend exploring the following:

          War on Ice

          Quant Hockey

          Sporting Charts

          Behind the Net

          Natural Stat Trick

          Stats.Hockey Analysis

          type any one into google and you’ll find them.

          BtN and HA aren’t as user-friendly as some, but both have some terrific data.

          Sporting Charts has the most visually-oriented display of the bunch, although WoI has some nice charts to help illustrate certain points.

  • prendrefeu

    Nice post Rex, great commentary Wolf!

    I agree with Jeff’s post (#9), I’m just generally super stoked to watch this team play, develop, and perform no matter what. Playoffs? Sure, that’s a nice bonus but that’s not the point. The point is development both on an individual level (so far so good among nearly all players) and as a team (without a doubt).

    I’m hoping that – without the reason being injury replacement – we get to see Poirier at the NHL level some point this season. What would a line of Gaudreau-Hudler-Poirier look like? What about Bennett?

    Just some development on D and I think this team is very close to being right up there with the top tier teams with a sustainable system to build off of. Heck, I even think Smid might be a keeper (low-pairing), he just needs a good partner to work with and he’s getting slightly better over time.

    Win or lose, it’s all very exciting to see how this plays out, isn’t it?

  • RedMan

    Has anyone read the Lambert article on Puck-Daddy? I think his entire agenda is just to write articles about the Flames, and how they are doomed to lose. I find it ridiculous and incredibly annoying that all he can do is recycle material he has written 50 times this season.

  • Section205

    per fenwick-stats.com:

    last 10 games 49.1% EV Fenwick close…
    last 23 games 48.9% EV Fenwick close

    These are decent numbers. At this level, you can accept that two good goaltenders and a few skill guys like Hudler, Gaudreau, Monahan, Gio and Brodie (good powerplay) can compensate for a small shortfall in unblocked shot attempts and generate a decent winning percentage.

    First 5 games were nasty, but since then we’ve been fine, possession-wise.

    Obviously our SH% is high and we might not expect this many goals all year long. I don’t think we should expect a goal differential of +51 after 82 games.

    I would be thrilled with +25 and a playoff spot, for this year. Even with 49% EV Fenwick Close the rest of the year. We can build on that next year.

  • mattyc

    @ the wolf and others:

    While there are certainly folks who just expound Corsi gospel, any good scientist/analyst is constantly skeptical, trying to improve their theories. Having said that, this isn’t the first time a team has ‘defied’ their possession stats. There problem is that there is a mountain of historical data that suggets that it isn’t sustainable. To suggest that this iteration of the flames is somehow different from those other historical cases requires some sort of proof. Extraordinary claims require extraordinary proof. IMO there just isn’t any data that supports it.

    There’s lots of good signs with the flames (giordano, Monahan, Brodie, gaudreau) and they aren’t bottom of the league bad, but I think it’s a tad optimistic to expect things like hudler burying his chances at double his career rate.

  • mk

    Random note of the day: Jarome Iginla was certainly an elite player in his prime – he has played 1337 NHL regular season games at this point… 🙂 Great number, eh?

    Anyone who trashes this doesn’t get the joke…

  • The Last Big Bear

    1) When I read your PDO table, i see 23 out of 29 teams (that’s 79%) with “unsustainalble” PDOs who either stayed about the same (within 2 or 3 places in the standings), or actually went significantly UP in the standings the following season.

    Teams which finish with a PDO of more than 102 are more likely to finish top-2 in their conference the following year than they are to drop off in the standings. A high PDO appears to be strongly predictive of future success, regardless of whether the PDO regresses subsequently.

    2) Of course the teams with the highest PDOs see a reduction in PDO the following season. What possible alternative could you expect to see? The top teams continuing to further progress in an infinite cycle of year-on-year inprovement? Hey, get this, on average, the Stanley Cup winner actually FALLS in the standings the following year. Rarely does the 1st place team finish any further ahead the following season. Similarly, the Art Ross winner usually has fewer points the following season.

    3) Your weather analogy has finally and completely convinced me that PDO is completely useless.

    In your scenario, what role does the thermometer have? A warm temperature in December is disregarded because we know its winter, and a warm temp in June is superfluous because we know its summer. The thermometer reading itself contributes nothing, becsyse you are basing your weather predictions solely and entirely on outside information.

    You are saying the temperature will regress, not because of anything the thermometer says, but because you think it is December. People are saying the Flames will regress, not because of their PDO, but because they think the Flames are bad.

    I’m not even saying they’re wrong, or that the Flames won’t regress. I’m just saying that the Flames PDO adds nothing to the conversation.

  • RexLibris

    Why does no one look at the Anaheim Ducks and say they must regress? I realize, Rex, this has nothing to do with your article, still, the Ducks are sitting with the 2nd best winning percentage in the league, while their CF% has them ranked 20th overall and treading just above water at 50.1%. Yet, nary a bad word about this team and how “lucky” they are.

  • The Last Big Bear

    To avoid coming across as overly negative, I feel I should point out that I think the percentages chart was a solid contribution. It contains much of the exact kind of information I want to see to make decisions about “sustainability”.

    And on the balance, any article that contains Dire Straits clips will get a thumbs up.

    • RexLibris

      The teams that maintained a high PDO were ones that could boast elite-level skill and thus retain a higher sh% or sv%.

      If the Flames had added elite-level skill during the off-season, and were getting a sh% around 14% from a wider variety of forwards then it is likely that their PDO would normalize and the successes would be less conspicuous.

      Instead what we see is a handful of forwards with +20% sh%, Hiller having a strong season and Ramo floating nearly Vezina-esque numbers.

      The weather analogy works because you are basing your perception of the temperature relative to the norm. You must look at things within their context otherwise you are focusing on too narrow an experience and lose perspective.

      This is the gist of Kent’s favourite saying: can’t see the forest for the trees.

      To take your seasonal reference and re-apply it to hockey, the teams that have a high PDO and find themselves regularly in the hunt for a championship are not being floated by something unsustainable. They are legitimately earning their success because they have good players who can do this consistently.

      The teams that have a high PDO and find themselves outperforming objective expectations are more often than not achieving success despite a mediocre and/or young roster. This isn’t to say that the talent isn’t there somewhere, but it may not yet be fully matured or well-balanced enough to sustain that level of performance. Colorado is an example of this. Perhaps the Flames will be so as well.

      Time will tell.

      Thanks for mentioning the DS clip. I think you were the first. 🙂

      • The Last Big Bear

        You are just further reinforcing my point.

        Your perceptions of what is sustainable or not, or how a team will perform moving forward, is based on whether a team is good, whether they are a consistent playoff contender, etc. A high or low PDO doesn’t change your opinion one way or the other.

        Let’s break it down into logic tables:

        Good team, high PDO? That’s sustainable -> good team will be good

        Good team, bad PDO? Due for upward regression -> good team will be good

        On the other hand:

        Bad team, good PDO? Due for downward regression -> bad team will be bad

        Bad team, bad PDO? That’s sustainable -> bad team will be bad

        In all cases, we project that good teams will be good, and bad teams will be bad. PDO contributes nothing to our assessment. It is entirely a red herring.

        • RexLibris

          PDO is useful in helping to separate the narrative from the facts.

          Eg: the lockout season where the Leafs made that run into the playoffs.

          Fans went nuts, the media bought into the hype and perpetuated it, objective observers who looked at the numbers said it wouldn’t last.

          Sure enough, the following season the Leafs crater, finishing 12th in the Eastern Conference and 23rd overall.

          How did this happen? people asked.

          Because they were extraordinarily lucky and some people need to be shown hard data to accept that their folk heroes aren’t everything that the fans make them out to be.

          PDO isn’t useless, it is just misunderstood and often applied too liberally.