The Vancouver Canucks Have Been Lucky

B3uQtCICYAAI-HX
Courtesy: @VanCanucks

Luck plays a massive role in determining the outcome of National Hockey League games, mostly due to the parity, introduced by the salary cap, and because of the relatively low number of events per game.

When we talk about luck within the analytics circles it is often assumed we are referring to PDO. This is
normally true, but PDO does not capture all portions of results that are outside of players’ control.  So far the Canucks have a fairly normal PDO
but they still have been receiving the benefits of luck other teams have not.

Read past
the jump if you’re feeling lucky.

Rob Vollman
wrote about luck in his
2013 Hockey Abstract. 
He has a really nifty tool on his website, the “
Hockey Abstract Luck Chart,” which shows how lucky teams have been over varying attributes.  These includes: PDO, the summary of shooting
and save percentages which regress to the mean; special-teams index, a “PDO” for
your power-play and penalty kill; teams’ success in 1-goal and over-time games; and a
teams’ CHIP, or Cap Hit of Injured Players.

Injuries
are a form of luck.  Every team should expect to experience some form of injuries within their roster throughout
the season.  Teams who experience a large
number of injuries, from a number of players, especially those major injuries to the top end
players are considered “unlucky” while the opposite holds as well.

@LW3H keeps track of all teams’ CHIP numbers and
updates them throughout the year.  His
most recent report of all teams through the first 20
games was just released today.  The idea behind
CHIP is that typically teams pay the better players a higher salary compared to
replacement level players.  Comparatively
a team losing a first-line player (like a Dan Hamhuis) is more likely to be severely impacted than if they’ve lost a
fourth-line player (like Tom Sestito).  CHIP adds up all the
salaries-per-game for all of the injured players on a team, the higher your
CHIP the more injuries to better players and the harder it becomes for the team
to win.

So how have
the Canucks fared so far?

CHIP graph (20)


That’s
right, the Canucks are right at the bottom of the league in terms of injuries.  All of the numbers can be broken down further
here.  In
simple English, relative to the rest of the league, the Vancouver Canucks have
had almost no injuries to key players which likely has played a role in the Canucks currently sitting at near the top of the Western Conference. 
The “worst” injury the Canucks have had to deal with so far is the loss
of Radim Vrbata for a few games.

Now this is
not to downplay the success of the Canucks, because they have played better
than expected; rather, the teams the Canucks are competing against have not
had the benefit of equal luck.  Within the Pacific
Division the real race for playoff positions are against: Los Angeles, San Jose
and Anaheim.  By Score-Adjusted
Corsi
these teams
are all better than the Canucks.  When
looking at their PDOs two of these teams have been unluckier, and when we compare their CHIP to the Canucks they have all fared worse off.

We can use
a simple Monte Carlo simulation to try and predict the rest of the season and there’s
no surprise that all three of these teams are likely to vault over the Canucks
over the next 60 or so games.  In the end
the most likely standings within the Pacific Division, based on the teams as
they are right now, are: Anaheim, San Jose, Vancouver, Los Angeles, Calgary,
Arizona and Edmonton.

1JzfZWj


As the
current CHIP report is only through the first twenty games it does not yet
capture the effect on the Canucks of having lost Hamhuis.  I imagine by the next CHIP report the Canucks
will fall in the rankings closer to the middle of the group, but it’s unlikely the Canucks will see the
injuries that the Columbus Blue Jackets have faced.

Luck matters in hockey. Using data like PDO we can forecast the
direction a team is likely moving in as regression has enough time to take hold.  Injuries are another form of
luck, but it’s nearly impossible to predict how many times
the injury-bug will strike, and for how long players will be out.  There is work within academia and other major sports, such as soccer, that uses biometrics to try and predict when over-usage injuries are likely to occur.  Finding ways to forecast these will greatly give teams another competitive advantage.

  • prendrefeu

    So they are also projected to finish ahead of LA in the pacific and make playoffs?

    Many fans including myself would consider that a success.

    While the work is appreciated I dont think many people think they will finish on top in the pacific division anyway. Statistics or not, to many good teams.

  • prendrefeu

    Take a look at last year’s final league standings and rank all 30 teams based on wins. Boston with 54 wins ranks 1st buffalo with 21 wins ranks 30th. Now rank all 30 teams (1-30) based on 5-on-5 CF% and rank all 30 teams (1-30) based on 5-on-5 CF% in close game situations and compare those ranks to the win ranking.

    5-on-5 CF% ranking predicted the final standings (within +/-2 positions) for 6 of 30 teams.

    5-on-5 CF% in close game situations predicted the final win ranking (within +/-2 positions) for 8 of 30 teams.

    There is absolutely ZERO guarantee that a team with a higher CF% will finish higher in the standings than a team with a lower CF%.

    Injuries are way too random to try to quantify. 9 times out of 10 Hamius misses two weeks with a minor groin strain. It was a total fluke that he happened to blow out his groin completely. Trying to justify a teams success in the long term using something as random as injuries is….. I don’t even know the right words to describe it.

    This whole blog post just seems to be grasping at straws, clinging to the notion that the Canucks are a truly bad team and are lucky to be where they are because their injury luck hasn’t caught up to their middling CF% yet.

  • Cale

    “Within the Pacific Division the real race for playoff positions are against: Los Angeles, San Jose and Anaheim. By Score-Adjusted Corsi these teams are all better than the Canucks.”

    The Canucks are 9th and basically dead even with the Ducks in CF% with the score tied.

    http://stats.hockeyanalysis.com/teamstats.php?disp=1&db=201415&sit=5v5tied&sort=CFPCT&sortdir=DESC

    A deeper look at some of the possession stats shows where the Canucks are “struggling”…

    While the Canucks are 15th in Down1 CF%, part of the issue is that they are 22nd in the NHL in time spent down by a goal

    http://stats.hockeyanalysis.com/teamstats.php?disp=1&db=201415&sit=5v5down1&sort=toi&sortdir=DESC

    If they had spent more time down by a goal, which is counterintuitive, the possession numbers would be a little better but the W-L record would likely be worse.

    The Canucks are 21st in Up1 CF% and this will surely have to improve.

    But part of the issue is that Canucks have spent more time than any other Western Conference team up by a goal…which is actually a good thing since the objective is to win on the goal scoreboard not the possession scoreboard…

    http://stats.hockeyanalysis.com/teamstats.php?disp=1&db=201415&sit=5v5up1&sort=toi&sortdir=DESC

    What is also indisputable is that the Canucks and Sharks have played a disproportionate number of games on the road.

    We can say with absolute certainty that this will even out…

    I’d suggest the Canucks are a little better than their “middling” possession numbers but not quite as good as their W-L record.

    The most interesting thing is that the Pacific division looks more open this year while the Central looks more like the Western powerhouse.

    The non-delusional amongst us would do well to acknowledge this…

  • prendrefeu

    The problem with this article is that it assumes all luck will wash out over the course of a season. It clearly does not. It also assumes that injuries and the outcomes of one goal games are entirely random. While PDO has been studied at length, CHIP and the other figures have not. I have serious doubts that there are not some systematic factors at play. Also they may not regress to the mean at the team level.

    A lot of sloppy assumptions make this piece a real mess.

  • prendrefeu

    @JoshW

    “The “worst” injury the Canucks have had to deal with so far is the loss of Radim Vrbata for a few games.”

    Not a few, just a couple, as in 2 games. Hamhuis is far worse when you consider duration.

    Also Luongo, does the $800,000 per we’re paying him count for our CHIP stats? Because they should by these metrics we’re paying him and he’s not playing, he’s actually injured at the moment. Where do retained salaries factor into the equation?

    “Within the Pacific Division the real race for playoff positions are against: Los Angeles, San Jose and Anaheim.”

    If we pretend Calgary isn’t right there playing good hockey they’ll go away right? You do realize San Jose is 6 points behind them in the standings right now. Unless you think San Jose’s PDO regression will make up that 6 points, I think Calgary is part of the “real race” in the Pacific as their PDO is sustainable at 100.7

      • Cale

        My bad, then where does missing Burr for 5 games factor in? Based on time missed and salary I’d think that would be more substantial than Vrbata for 2.

        Ah screw it, it’s just bad analysis. Too much cherry picking for the sake of an article. I’d argue losing your best shot blocker or underpaid defensive specialist/shutdown guy might have a bigger impact on luck than, say, an overpaid non producing winger (not saying that Burr is one). Some of these stats are useless attempts at explaining the random nature of the game. I definitely would rank CHIPS up there with the more baseless of them.

  • Cale

    CHIP is interesting, neat to see the attempt is being made to quantify and analyse this.

    It strikes me as a bit flimsy though, just sort of lumping the total cap hit together seems a bit ham-fisted beyond providing a very broad overview, I don’t really see how you can conclude much from it.

    For one thing I question the correlation between cap hit and real value to the team. I think there are too many preconceptions (for example all the old scouting attitudes the new analytics have been calling into question) and market forces at play in driving salaries for us to really count on that as an indicator of real value in any sense but the very broadest.

    Also, while I like that it’s broken down by position on the chart, I wonder if some weighting of the various positions might be called for. For example is losing Dan Hamhuis going to hurt the Canucks more or less than if the Flames lost Hiller? Or are the salaries of scoring forwards (even the very inconsistent ones) not generally higher than the quietly effective linchpin defenders like Hamhuis?

    It’s a very interesting direction to look in, but isn’t a pretty shaky foundation for building any kind of conclusions?