Icings, The Most Advanced Stat of All



In recent years, there’s been a lot of discussion about advanced stats. We talk about PDO, Fenwick and Corsi as ways of determining how teams and individuals are playing, who’s driving play and basically trying to find handy ways of evaluating hockey players – which, as we’ve discussed a lot here, can be hard.

Since last season, though, I’ve begun to think that one stat has been overlooked – icing differential.


Icings often occur because a team has gotten themselves so hopelessly hemmed into their own zone that they have no other choice but get scored on or throw the puck donw the ice. The Flames iced the puck a lot last year. They also got killed possession-wise. These two things may be connected.


Through three games this year, the common perception is that the Flames generally carried the play against their opposition. How did they stack up in terms of icings?

Washington 4, Calgary 4: Most of Calgary’s icings happened in the second period, while Washington iced the puck throughout the first and third.

Columbus 5, Calgary 0: In a rare icing shutout, the Flames went the entire game without getting flustered enough to ice the puck. Columbus iced the puck fairly evenly throughout the evening.

Vancouver 8, Calgary 2: Outside of a single shift where Calgary iced the puck twice in succession, the Flames generally dominated possession and icings.


The team’s leaders in icing plus/minus are probably the players you’d expect. Dennis Wideman, who’s been strong to start off the year, is +7 – meaning he’s produced seven more icings than he’s been on for.
Wideman’s defense partner Kris Russell is among four tied at +6, along with T.J. Brodie, David Jones and Ben Street. Street’s performance is the most impressive when you factor in that he’s played significantly less (36:33) than the other three (between 54:42 and 74:13).

Shane O’Brien is the only player in the red for icings (-2), although Matt Stajan (who only played one game) and Chris Butler, O’Brien’s partner, are even for icings.

The Flames, summarized: Wideman (+7), Street (+6), Jones (+6), Russell (+6), Brodie (+6), Glencross (+5), Giordano (+5), Galiardi (+3), Bouma (+2), Monahan (+2), Stempniak (+2), Backlund (+2), Hudler (+2), McGrattan (+1), Colborne (+1), Baertschi (+1), Stajan (E), Butler (E), O’Brien (-2).


So what do you think? Are icings a decent metric? Does this measure need to be tweaked to account for ice-time or quality of competition? I like the metric because it’s a counting stat that doesn’t rely on scoring. Can it be improved?

Fire away in the comments.

  • Interesting Ryan.

    Is there anywhere else this is kept, for other teams? I say wait and see. If the metric aligns well with wins and losses, how the Flames look to the eye, and the standings, you may be onto something here.

    The most novel part is it already adjusts out the PK.

  • piscera.infada

    It is interesting. Sounds like something you would need to have hand and hand with zone starts though. O’Brien for example (worst on the list) had highest d-zone starts (35.1%) and the lowest o-zone starts (18.9%)) of all Calgary D.

    Meanwhile Wideman (best on list) was the exact opposition (o-zone 51%, d-zone 19.6%).

    I think it is more meaningful at a team level. Or for players like Brodie, Glencross, Hudler, etc that saw a relatively even distribution of offensive/defensive zone starts.

  • Icing is not necessarily about playing desperately while being hemmed in your own zone. With the two-line pass gone, often icing is a result of trying to create a breakaway for an open teammate that goes slightly awry. I certainly wouldn’t count icing as a meaningful statistic.

  • Like all stats, there’s going to be some noise. Hell, even if you see a guy with a goal on the board, you don’t particularly know if he deked the entire opposition out to score or the puck just happened to glance off his leg in front of the net.

    For icing differential, you’d have to get a big sample size, correlate it other meaningful measures and see if it adds anything useful or not.

  • NHL93

    Very interesting.. though doesn’t a team defending a lead late in a game tend to ice the puck also? If that can be factored in, the metric can say some interesting things.

  • NHL93

    I like the idea. I have been troubled with the Flames recent past in which it seems, and I don’t have the stats on it, that their opponent scored shortly after the puck was iced because a tired line couldn’t get off the ice and regain possession.

    You could, if it were useful, broaden the notion to “adverse possession events” that would include those troublesome delay of game calls that occur when a player is merely trying to clear the defensive zone and/or ice the puck and misses.

  • ChinookArchYYC

    This is a really good idea. It’s doubtful an Icing Stat could stand on its own, as a stat that measures performance about any one player or team , but in conjunction with other stats like the aforementioned Corsi, I can see this being very useful.

    In saying that, it would be interesting to see what an Icing WOWY would produce for defensive pairings and their relative time on ice. Icing WOWY/60 in 5on5 play might yield some quantitative ways to measure how good defensively a D-pairing is. And finally a metric for defensive-minded pairings . . . Well okay maybe.

  • NHL93

    i like it, and it seems like a good way to evaluate games.

    problem you’ll run into when you try to draw more heavy conclusions is one of the same that applies to goals… i.e. relatively small number of events.

  • Derzie

    If the icings are lopsided, and the score is the opposite that would indicate non-average goaltending. Either really bad, really good or both. For instance the Vancouver stat indicates that Joey McB was pretty bad, Lack was pretty good or both. Also, if you are on the PK most of the game your icings will be few so that needs to factor in somehow.

  • The Last Big Bear

    You’re forgetting the Nations’ mantra:

    “If it doesn’t show up in shot differential, it’s not real”

    As such, I refuse to accept this reasonably compelling heresy.

  • The Last Big Bear

    Also, it needs an obscure name.

    Just as “goal differential” has to be called “+/-“, and “shot differential” has to be called “Corsi”, and “shooting% plus save%” has to be called “PDO” even though PDO doesn’t even stand for anything, so icing differential needs a stupid and confusing name.

    I vote we call it “A-stat”.

    ie “The Flames appearance of a strong start to the season is supported by their strong A-stat.”

    We have to make sure that these stats are not readily accessible to the casual fan. Otherwise we won’t be able to get away with calling them ‘advanced’ any more.

  • beloch

    It’s interesting to track a bunch of stats that might be correlated to possession time in the offensive/defensive zones, but why not just track those directly? Take two stopwatches to the game: one for each zone. Start the appropriate timer when the puck passes the blueline. Done.

  • beloch

    Pretty interesting point and there might be something there, but I think it might be taking advanced stats a little too far, for a couple points

    1. I think with the amount of average icings per game quickly from this article: 7.5 (albeit only three games), there may too many unconnected stretch passes that don’t really make it a good indication of being hemmed in. I could be wrong about that though.

    2. The number of icings may also be inflated when a line is trying to get a change but ends up icing it, and then quickly ices it after due to tiredness or whatever. I guess maybe that’s what your trying to study, but I think it’s just a little too much to really tell you anything about the performance of a team or player.

    Interesting though

  • Icings are a control tempo part of a game. From line match up which helps centers face off pct by facing a “usually tired line”, to the overwhelmingly increased fenwick. Icings matter. You travel less to get to a high scoring control-tempo area. Which, to touch on Kent’s point will increase your offense by “creating a modifier influencing possession, like zone starts or quality of competition.”. This is where the game is won and lost when bad icings kill good team stats on zs and pdo which create more chance for the opposition. Limit opposing offense thru a offense which drives play and rarely