Calgary Flames Adjusted 2013 Possession Rates



pic via Shane Henderson

Another year is over and we’re left to pick through the rubble a bit. The first order of business is evaluating the various skaters by possession – or the ability to drive play, which is one of the metrics in hockey that has a strong signal-to-noise ratio.

That said, we can correct corsi for a number of factors be better put it into context and up the level of confidence with which we can say the number is indicative of the player’s ability and not some other factor (be it luck or circumstances). Two major issues with raw possession rate is zone start (or how often a player starts his shifts in the offensive zone) and quality linemates and opposition (playing with or against good/bad players).  

Good news is, Oiler fan and mathy blogger Michael Parkatti came up with a regression equation that helps correct for both issues recently.

The formula would then be:

Expected raw Corsi = -11.91 + QualDiff * 1.00 + Off ZS% * 0.24

So, a player with QualDiff of 0, and a zone start rate of 50% would have an expected Raw Corsi of:

Expected raw Corsi = -11.91 + 0 * 1.00 + 50 * 0.24 = -0.15

The formula suggests that Corsi will increase by 1 for every unit increase in his quality of teammates and every unit decrease in his quality of competition. His Corsi will increase by one about every 4 more percentage points in his zone start. A player with a zero QualDiff and starting every faceoff in the defensive end is expected to have a -11.91 Corsi, reflected in the intercept above.

So that’s a lot of numbers and formulas and such, but what it gives us is a really easy way to plug in raw corsi rates and adjust for a players circumstances. That means it’s easier to compare guys on the same club across various lines and roles.

To the tables!

The Forwards 

Player Corsi QoC Corsi Qot Qualdif ZS % corsi expected corsi on corsi adjusted
Backlund 0.52 -4.98 -5.50 45.40 -6.52 2.14 8.66
Stempniak 1.20 -4.39 -5.59 43.50 -7.06 -0.55 6.51
Stajan 1.26 -4.29 -5.55 42.00 -7.38 -3.13 4.25
Glencross 0.50 -3.85 -4.35 49.30 -4.43 -1.17 3.26
Cammalleri -0.01 -4.79 -4.78 45.40 -5.80 -3.17 2.63
Begin -1.84 -6.49 -4.65 46.30 -5.45 -4.01 1.44
Jackman -1.81 -6.61 -4.80 46.90 -5.45 -4.38 1.07
Hudler 0.03 -4.69 -4.72 44.10 -6.04 -5.32 0.72
Tanguay 0.95 -3.73 -4.68 45.20 -5.74 -5.12 0.62
Horak 0.52 -6.32 -6.84 39.20 -9.35 -9.65 -0.30
Cervenka -0.68 -4.85 -4.17 43.80 -5.56 -6.05 -0.49
Jones -0.89 -6.81 -5.92 45.80 -6.83 -8.78 -1.95
Baertschi -0.93 -4.78 -3.85 48.20 -4.19 -6.99 -2.80
Reinhart -0.28 -5.31 -5.03 40.00 -7.34 -14.39 -7.05
McGrattan -2.56 -6.90 -4.33 57.10 -2.54 -19.74 -17.20

The first three columns have to do with each forwards’ quality of competition and linemates, as well as the "difference" between them (Qualdif). Next, we have each guys zone start ratio (offensive zone/defensive zone) and his resultant expected corsi based on Parkatti’s equation.

Corsi on is each skaters’ actual corsi/60 rate (via and adjusted corsi if the difference between his actual and expected rate – meaning a number above zero indicates a guy outperformed his expected rate according to his circumstance (that’s good!) while a negative number means he underperformed (that’s bad).

The table is sorted by adjusted corsi.

Of course, none of the results should surprise anyone, but they are interesting nonetheless. The guys who looked good by eye and by underlying numbers come out well here: Backlund, Stempniak, Stajan and Glencross lead the way. Jackman and Begin come out well for minimum wage fourth liners too.

Roman Horak actually moves way up with this method thanks to rough circumstances, but doesn’t quite come out even. Still, not a terrible result for a kid. Baertschi isn’t quite there, but his improvement was rapid upon his recall and remember we’re talking about pretty small numbers here.

Reinhart, on the other hand, is further behind, especially because he doesn’t bring the same offense or PP prowess as  Sven.

Finally, Brian McGrattan is obviously an enforcer, with typical enforcer type numbers.

The Defense

Next up, the blueline:

Player Corsi QoC Corsi Qot Qualdif ZS % corsi expected corsi on corsi adjusted
Brodie -0.42 -4.87 -4.45 47.90 -4.87 0.97 5.84
Wideman 0.50 -4.56 -5.06 45.50 -6.05 -2.98 3.07
Sarich -1.97 -5.83 -3.86 41.20 -5.88 -3.28 2.60
Giordano 1.30 -4.36 -5.66 41.00 -7.73 -5.96 1.77
Smith -0.68 -5.31 -4.64 53.30 -3.75 -4.62 -0.87
Carson -1.40 -7.04 -5.64 63.60 -2.28 -6.21 -3.93
Butler -0.12 -5.20 -5.08 46.60 -5.81 -13.50 -7.70

Again, no big surprises, except for maybe Cory Sarich who despite his various limitations never has terrible underlying numbers. Brodie and Wideman lead the way with the kid out in front. Giordano also is somewhat vindicated by this process thanks to the high difficulty of his role this year.

On the other hand, Carson and especially Butler don’t really look like NHLers. Butler is out on an island of suck all by himself, in fact. That might have been understandable with the dude trying to play over his head last season, but this year he was mostly a third pairing guy and still couldn’t even come close to breaking even.


The best news for the Flames org is their two main youngsters are already the best possession players on the team and the Flames have the option to sign them both up for cheap this summer. Niether Backlund nor Brodie is a superstar in the making, but they’re both already better than your average NHLer, are young enough that they should improve, and will not cost an arm or leg to re-sign.

Caveat emptor; we’re talking about a small sample of games here in general thanks to the shortened season, and even smaller in individual cases thanks to injury and other factors (like playing between the AHL and NHL for the kids). In addition, with the team trying to lose down the stretch by throwing every non-NHL body they could on the ice (and Hartley not matching lines or anything), some of this sample isn’t exactly representative. So keep thosee factors in mind. 

  • We’d probably have to re-run numbers for the current season as well as a full year to get a good sense of things. The table you are referring to was compiled in the middle of March, so we’re talking a 25 or so game sample.

  • seve927

    So following the link to the fan blog – I had a look at the Oiler stats. Why are they all showing such great adjusted corsi? Hall, Eberle and Nugent-Hopkins are all given about an 8 point boost because of crappy teammates? Each other? I don’t understand.

      • seve927

        But it just doesn’t seem to mesh with looking at the WOWY numbers. RNH spends most of his time with Hall and Eberle, and he’s good with them. What does the rest of the team have to do with his on-ice numbers. How are guys he’s not playing with dragging him down so much, as would be implied by these numbers? And some of the guys that are dragging him down (Yakupov, Smyth, Paajarvi) are presumed to be dragged down by their poor teammates, and would actually be great otherwise.

        Something in this analysis just doesn’t seem right to me. But I’ll admit I may be in over my head πŸ˜‰

        • SmellOfVictory

          Feels off to me, as well. It looks like the calculation might have a bias toward players who are substantially different from their teams (e.g. good players on awful teams, or conversely bad players on really good teams). Especially when you look at the top 10 and see guys like Brandon Dubinsky, Andrew Cogliano, and Patric Hornqvist. Actually, not a single guy in the top 10 is playing on a good possession team.

          I guess that answers my previous question as to whether or not this was useful for comparing across teams (the answer is a clear “no”).

        • Danger

          Just a guess, not having looked at the numbers extensively, but I’d think the kids’ numbers are hurt not by each other but by the D who are on the ice at the same time as them. Since Edmonton’s D corps is truly abysmal, it makes sense that they would manage to bring down the corsi for every good forward (and maybe even some of the not so good ones). At least, that’s how I understand it – I’ll admit that I too might be in over my head here πŸ™‚

          • seve927

            Yeah, I would have said so as well, but still just looking at RNH numbers (don’t want to spend too much time on this), his numbers still seem good with the defensemen he’s on with the most. Most of his time was with the Schultz’s and he was better than 50% with both of them. A little worse for Smid and Petry, but not much. It just doesn’t add up to me.

  • Jeff Lebowski

    If I understand this right, it would be nice to see a moving adjusted corsi to see how players are trending. However I realize that this specific example is a year end review.

    As with Baertschi the overall number may be negative but he was climbing out of a hole. Without that context it would be easy to assume he is just not a good player yet.

    If it were last 10 games vs season (or last season or last 10 from last season) it would be interesting to see how performance drops or rises and why (is there a brutual road trip with lots of travel), are there teams that players play significantly good/bad against? Does this help with matchups? Is Butler terrible against teams who are big and favour a dump chase or is he any better against smaller lineups that play puck possession? Should the coaches tinker with the lineup accordingly?

    I dunno just a thought.

  • Derzie

    This stat collection is a ‘real-world’ beauty. Watching this team every game this year, the order matches what I observed. Great numbers. I would like to see the numbers compared between teams at each end of the spectrum. In other words, how many players are below, at or above NHL standards. Given our results, a lot more below. Is Backlund an average NHL’er? Brodie? Again, great stuff.

    • SmellOfVictory

      I personally think what you’d get, if you put all ~600 skaters on a spreadsheet with their adjusted Corsi rates, is a median of zero. If it’s not zero, or very close, then someone probably screwed up somewhere in the formulas.

      So while we don’t have enough information to say to what magnitude a given player is playing above or below average in possession terms, I’d say an easy rule-of-thumb would be that positive is above average and negative is below average.

  • Colin.S

    So does this mean the Wideman signing may not be as bad as most people made it out to be. I mean it still has the potential to be bad, but it looks like he was able to be a positive contributor on the team.

    He led the blue line in points, TOI(average for the year, Brodie was getting more minutes late in the year) and had the second best adjusted Corsi, so if he is able to do that again over a full season then maybe Feaster did right by signing him, maybe?

    • I was never down on the Wideman contract because of the player necessarily – only that the team assumed a lot of risk over the life of the contract by giving him tern, dollars and a NTC. He was a decent player for the team this season and probably will be for a year or two more, but there’s a chance it goes south before the thing ends.

  • mattyc

    Maybe I’m mistaken, but wasn’t Jones a pretty good possession player last year?

    Would be interesting to see where pre-deadline Iggy and JBo fall too.

    BTW is there an easy way to separate out team or player stats based on sections of the season. For instance if I want to see how the team’s PDO has been since April 1, would I have to manually put it together, or has someone done all the coding?

  • SmellOfVictory

    OOoooOoo this is a new one. Would it be inaccurate to say that this stat is almost good enough to use across teams as well as within them? It’s basically adjusting for overall team effects, afterall.

  • TheRealPoc

    Words cannot express how happy I am that the underlying numbers are corroborating with my eye test on Butler.

    I mean, I’m not “happy,” because he’s a Flames roster player…but, y’know…yeah.