Scott Arniel lost his job as Columbus head coach this week. The only thing that was surprising about the move was that it took so long: as many have pointed out, the move comes far too late to save the Blue Jackets' season, as they already sit an insurmountable 21 points out of a playoff spot. What is also not surprising, as my colleague Robert Vollman has pointed out to me, is that it follows a pattern of coaching firings in the NHL in recent years: fire the coach when the percentages (a.k.a. "the bounces") are playing against the team; things will eventually go the team's way again, justifying the firing and making the GM look smart in his assessment that the previous coach had "lost the room".
Indeed, some of the coaches who lost their jobs this season, like Davis Payne, Jacques Martin, Bruce Boudreau, and Terry Murray, were coaching average-to-good teams that were underperforming. All four teams had positive or even Corsi and were getting out-percentaged by the opposition. The Blue Jackets seem to follow this trend: their overall shot differential this season is positive, 1199-1095, despite the fact that they've been outscored by nearly 30 goals. Clearly it's just a matter of time before things bounce back in their direction, right?
Not so fast. Here, for your appraisal, is the table of NHL teams' goals and shots over the last three seasons (2009-10, 2010-11 and the half of 2011-12 so far), ranked by goal differential:
Team GF GA SF SA GD GD (shots) GD (PDO)
Vancouver Canucks 616 465 6247 5911 151 30 121
Boston Bruins 531 422 6306 6100 109 14 95
Chicago Blackhawks 604 501 6537 5424 103 92 11
Washington Capitals 593 499 6140 5804 94 33 61
San Jose Sharks 557 479 6466 5778 78 68 10
Philadelphia Flyers 579 511 6182 5697 68 43 25
Pittsburgh Penguins 556 489 6383 5564 67 71 -4
New York Rangers 519 455 5722 5818 64 -4 68
Detroit Red Wings 564 501 6617 5853 63 71 -8
Buffalo Sabres 538 500 6188 6093 38 6 32
St. Louis Blues 513 489 5891 5579 24 23 1
Los Angeles Kings 489 471 5840 5548 18 27 -9
Phoenix Coyotes 492 483 5985 6079 9 -19 28
Montreal Canadiens 492 484 5941 6073 8 -11 19
Calgary Flames 506 500 5819 5763 6 -1 7
Nashville Predators 491 489 5769 5958 2 -17 19
Dallas Stars 526 536 5726 6115 -10 -31 21
New Jersey Devils 443 472 5686 5285 -29 39 -68
Anaheim Mighty Ducks 519 553 5600 6355 -34 -65 31
Minnesota Wild 470 519 5318 6140 -49 -66 17
Tampa Bay Lightning 520 570 5868 5917 -50 -5 -45
Florida Panthers 467 518 5880 6455 -51 -47 -4
Winnipeg Jets 521 574 5994 6375 -53 -42 -11
Carolina Hurricanes 519 575 5918 6394 -56 -36 -20
Colorado Avalanche 521 582 5748 6208 -61 -40 -21
Toronto Maple Leafs 513 590 5982 6027 -77 0 -77
New York Islanders 497 592 5844 6203 -95 -29 -66
Ottawa Senators 478 575 5854 5951 -97 -18 -79
Columbus Blue Jackets 474 591 5912 5929 -117 2 -119
Edmonton Oilers 468 591 5286 6253 -123 -88 -35
This contains just the basics: goals for and against, shots for and against, goal differential, and the goal differential due to shots, which I have calculated as taking shots in every manpower situation and multiplying them by the average success rate in that situation across the NHL (7.9% for 5v5 shots, 12.0% for 5v4 shots, etc). Note that this table excludes empty-net situations and does not count shootout goals.
What is the takeaway from this table? In Columbus' case, it is that their recent case of PDO underperformance is not an isolated incident. Columbus has been outscored by 117 goals over two and a half seasons, entirely due to being dominated on finishing ability, not puck control. This is a result of having one of the weakest #1 goaltenders in the NHL, Steve Mason, as well as a weak forward corps. This underperformance has persisted under three head coaches, Ken Hitchcock, Claude Noel, and Arniel. What is even more surprising is that Columbus broke even on Corsi in a division with two of the three most dominant possession teams (Chicago and Detroit) in the league, but was abysmal in percentages against teams that didn't excel at them.
Could this be due to luck? Quite likely some of it is. Columbus has had the worst percentages in the league over this period, and when you're the worst on a noisy statistic, chances are that luck played against you over this period. The standard deviation of goal differential due to PDO was 51 goals. The estimated amount of noise on this is calculated as roughly sqrt(GF + GA), which is 32 goals. Thus, the standard deviation of skill is sqrt(51^2 - 32^2) = 40 goals. That's right, even after over 200 games, almost half of your shooting percentage and save percentage is luck. Note that this is roughly in line with what I found looking at three seasons of data in my goaltending analysis here. The standard deviation of goals due to shot differential over this period is 44 goals, so at the team level, slightly over half of team talent exhibits itself as puck possession, and slightly under half exhibits itself as finishing (and preventing finishing). It just so happens that the possession talent is much, much easier to measure.
However, this does mean that there is significant true talent in finishing ability, and if we look over a long enough window we can see it. It is not surprising to find that Vancouver and Boston are the best PDO teams in the league: both of these teams combine elite goaltending with significant forward talent, although it's not surprising that in Vancouver's case the outperformance is more heavily loaded towards offense (the Canucks' shooting percentage of 9.9% over this period is the best in the league) and for Boston it is more weighted towards defense (a save percentage of 93.1%, miles better than the second place Rangers with 92.2%). Vancouver's PDO is 1020 while Boston's is 1015. Also, for those who think this still might be a fluke, in 2008-09, the year before my sample, Boston had the league's best PDO with a phenomenal 1030, while the Canucks had a respectable 1016.
We know that such teams are possible: the Edmonton Oilers of the eighties maintained a PDO of 1042 over seven seasons, and more recently the presence of Dominik Hasek allowed the Buffalo Sabres to post a PDO of 1026 from 1993 to 1999. The Sedins are not as good as Gretzky and Kurri, and Tim Thomas is not Hasek, but in both cases these teams have significant talent that allows them to outperform on percentages.
The opposite is true for Columbus: you don't fall short 116 goals purely by bad luck, as this would be a 3.5 standard deviation event, having roughly 1 chance in 5000 of occurring by chance. I wish Todd Richards the best, but the fact is that he's been trusted with a team of decent players with below-average finishing talent at the NHL level. The nights where the bounces don't go the Blue Jackets' way are not over yet.
Tom Awad is an author of Hockey Prospectus.
You can contact Tom by clicking here or click here to see Tom's other articles.