Where hockey’s new numbers get it wrong

Like a lot of people, I’ve become increasingly concerned with the importance placed on several of these newfangled stats that take away from my enjoyment of the NHL game. Hockey was a better sport before all the bookkeepers came along and started siphoning out the fun by cataloguing everything and determining the best players in the game based on who scored goals and made assists.

‘Goals’ is a very poor stat that has been taken far too seriously by some of the newer analysts. You keep hearing people talk about 20-goal, 30-goal and 40-goal scorers as if every ten goals to reach some imaginary plateau is what separates players from others. But what about the players who don’t play to score goals in the first place?

Take a defenceman, for instance. Many defencemen, including the speedy and shifty ones, don’t play close enough to the opposing net to score a goal. The best goal scorers from the defensive position in these playoffs are Drew Doughty, P.K. Subban, and Jake Muzzin, all with five. If you’re a proponent of the ‘goal’ statistic, do you mean to say that Muzzin is just as vital to the success of the Kings as Doughty? Anybody who watches the games would be able to tell you that Doughty’s vital defensive contributions, that can’t be counted, are more important for this team up 2-0 in the Stanley Cup Finals.

Then there are forwards who pass the puck rather than shoot. Many players can play on my team despite having very limited goal totals. Joe Thornton, for instance, scored just 11 times last season, as did Mikko Koivu. They were both centremen on playoff teams, need I remind you, while league leader Alex Ovechkin and his 51 goals didn’t even make it.

Clearly, “goals” are not a perfect way to judge players, but what about teams, even? The Stanley Cup is being contested this week between Los Angeles and New York. The Kings were 25th in goals for this season, while the Rangers were 18th. Better, more offensive teams, like Chicago, Anaheim, Boston and Colorado, met early playoff demises, despite their “goals for” statistic telling us otherwise. Let us not forget the Presidents’ Trophy-winning Bruins, who were 3rd in the NHL in goals, were eliminated in the playoffs by 21st ranked Montreal, despite the Canadiens scoring nearly 50 fewer times.

It’s replete with examples and the evidence is clear: goals are an awful way to judge players and teams because the statistic lacks so much context. I long for the days when we could watch a game, and determine the winner based on how the teams played the game. Hopefully this desire to count “goals” is simply a trend, as these numbers can not tell me anything I don’t already know about the game of hockey.

  • BuffaloBillsOfHockey

    When I first started reading the article I couldn’t get a scene from The Matrix out of my head. It was the one where Cypher, over dinner at a restaurant within The Matrix with Agent Smith, is striking a deal to have have himself re-inserted into The Matrix in exchange for his betrayal of humanity and he says, “Ignorance is bliss…I don’t want to remember nothing. Nothing. You understand? And I want to be rich. You know, someone important. Like an actor.”

  • Which Simmons article did I miss which bashes some stat or another??

    But seriously, all stats today fully suck with what will be available in the next three years.

    Edit : Well, I missed the first link. After spending 20 minutes reading a pretty awesome rebuttal by Eric, I think they are just having a violent agreement, though Scott might not want to admit it.

  • John Matrix

    Wow, I think scott13 kinda missed the point regarding individual PDO. I agree that comparing Booth’s individual PDO with Bolland’s individual PDO is pretty fruitless but comparing Booth to Sestito or Burrows? I think its a reasonable luck metric in the context of teammate comparison. As for team PDO… if you have enough data, I’m inclined to agree with scott13. I mean, in the short run, its safer to assume its luck but a season or two with an elevated PDO? That would be a great team, not a lucky one, right? Like, I’m guessing Boston has had an elevated PDO for a couple years now but that would be due to talent not luck, right? What do you think Cam? I’d love to see an article from you discussing the most useful applications of PDO.

      • BuffaloBillsOfHockey

        Wow. That was FAST.

        Seriously though, nice job of summarizing PDO and it’s uses. I can see how a series of articles like this one could really be a great tool for people wanting to get into advanced stats or even just statistics in general.

        Except for maybe the one guy who commented on the linked article. It’s always surprising how many angry, ignorant jock stereotypes there are out there and how oh-so-desperate they to believe that the game they play is one of heroism and magic; where men are men and the boys are culled through violence.

        The saddest part is the hatred they react with when presented with data by someone endeavoring to detect patterns in the game, sometimes for no reason other than for the sake of doing so; their sad, small little egos can’t withstand the idea that their “heroic” contribution to their sport can actually be reduced to a series of learnable actions and proccesses determined by if-then scenarios and limiting deviations from those behaviors.

        Saddest of all is the tired, old cliche that they all seem to use: “Have you ever even PLAYED the game, nerd?”

          • BuffaloBillsOfHockey

            “troll,” with a lowercase t.

            *Yawn*. This troll’s boring antics are making me so sleepy. Same tired writing style. Same pathetic taunts. Same decrepit routine of propping his own comments while simultaneously trashing the ones calling his own posts out. Same old mask he carves for himself by creating new forum identities and hiding behind them.

            NM00 recently offered the troll a Dark Knight quote. Let me offer the miserable, little troll a challenge in the form of one of my own:

            “Just take off that mask and let him come find you.”

    • BuffaloBillsOfHockey

      Booth is an interesting example for this thread, and probably supports Scott’s critique somewhat. Booth’s PDO was about even this year, but there’s a weird split: very high save percentage “canceling” a low shooting percentage. One of Scott’s main points seemed to be that it would be more useful to look at the two stats separately, and to put them in the context of career norms — in this case, that would mean focusing on Booth’s shooting percentage this year, which was below his career average and almost five percentage points lower than his 31-goal season.

      I still like combining the two stats into PDO as a sort of shorthand, especially when looking at guys like Edler who suffered from lows at both ends. It’s not very useful in isolation. In Booth’s case, there’s a good chance he’ll end up with similar PDO next year by getting worse goaltending while he’s on the ice (just through regression, if not goalie quality) and a better shooting percentage — which would likely mean more goals, and a worse +/- (which only matters when debating with people who still value +/-).

  • @Doc,
    The real point is that Alex Edler’s PDO is a useful thing, as it is very low (.938).

    Where there is disagreement is that Erik is saying “That is very low, thus he was unlucky”

    Scott is saying “It is very low and much lower than his career average , thus he was unlucky”. He maintains that we should factor in baselines for players before looking at their luck. For example, Henrik has had many years in a row over 1.0 PDO, thus his recent year of 1.0 PDO would be considered lucky.

    Note: They were talking about teams, which have a lower variance, but exceptional player-cases are still significant, in my opinion.

  • NavyChief99

    Spot-on article. Numbers all over the place are ruining hockey and baseball, my favorite sports. Numbers coming out of digestive vents of basement-dwelling dweebs mean absolutely nothing because not a single one of them is predictive. You don’t have to provide 63 random, made-up statistics for me to understand the game, numb-nuts, because I just watched the whole thing, and after 35+ years of viewing games on TV and in person I feel I have an adequate grasp of the just-concluded proceedings to not require your services. (PS, as a new poster, I followed above request to “pretend Mom is looking over shoulder” resulting in this version, which is much milder than what she would have said)