A lot of the narratives following the Olympics have had time to marinate and become entrenched in our collective memory as "established fact." However, a number of these storylines are either false or misleading, and I do my best to highlight why that is so below.
One of the best things about the stat movement in hockey is that it led to the creation of websites like Behind the Net, Extra Skater, and Vic Ferrari's time on ice reports. These websites take the relatively stodgy data from the NHL's play-by-play reports and reconfigure it into outputs we're all familiar with: Corsi, Fenwick, zone start, quality of competition, variants of ice time, etc. The benefit of these reports is that we can check to see if our eyes led us astray as we watched the game. Players who we thought looked good should have good underlying metrics; if they don't, then we weren't paying close enough attention.
Unfortunately, no similar data reports came out of the 2014 Olympics. The IIHF published very minimal stats after each game (ice time, points), but nothing even remotely approaching the detailed information we've come to expect from Extra Skater after a hockey game. The consequence of this was that the narratives after each game were allowed to run free, unchecked by the data-driven analysis which is the chief counter to hockey-storytelling divorced from facts.
I decided to go back and take an in-depth look at one game: the men's semifinal match between the USA and Canada. Why? Because it was after this game that the biggest narratives came to the fore, not only in describing Canada's success but also the Americans' failure. In re-watching the game, I tracked a number of things: Corsi%, scoring chance plus minus (SC +/-), zone starts, and certain match-ups. I only tracked even strength situations (which was most of the game since there were few penalties). I figured that with this information, we'd finally be able to evaluate each team's performance without having to rely on our memories or visual perception. The chart for Team Canada is first:
I don't think there's anything too surprising here. The whole team was really, really good, and it certainly shows in both possession numbers and scoring chances. Specifically, the Keith-Weber pairing and Crosby-Kunitz-Bergeron line were just obscenely filthy. They led in scoring chances and were tops in possession but had the "toughest" zone starts on the team, hovering between 45-50%. At this point in time, Keith-Weber-Crosby-Kunitz-Bergeron is probably the best five-man hockey unit in the world.
The only other thing that stuck out to me was how the Toews-Carter-Marleau and Getzlaf-Benn-Perry lines struggled more than most thought. Toews is regularly touted as one of the top 3 centers in the league and he was the tournament MVP in 2010, but he was decidedly average (certainly below average by his standards) when Canada played the US this year. Likewise, the goal scored by the Getzlaf line probably masked the fact that they weren't controlling play at even strength. Canada was still an absolutely stellar team, and it certainly says something about their depth when their two "worst" lines are only drawing even in possession and scoring chances. That being said, it's not true that Canada was able to roll out four dominant lines during that game; only two of them were able to consistently beat the opposition.
Here is the same chart for Team USA:
[Note that Faulk and Wheeler only played 2-3 shifts, so we should probably disregard their numbers.]
Quite the opposite story (unfortunately) for the Americans. The whole team was drowning in terms of possession and scoring chances. The best d-pairing was Suter-McD, who put up respectable numbers given their brutal starts. But even adjusting for zone start, neither of them drew even in possession. Additionally, Orpik struggled a lot in this game, which was no doubt related to the loss of Paul Martin.
As I re-watched the game to track these events, I couldn't help but constantly think how overwhelmed nearly all of our forwards looked. They struggled mightily with winning battles along the boards, and it seemed like they couldn't pass or receive passes that well, which made it difficult to establish possession in the offensive zone. These numbers tend to verify that, as only Kessel was driving play (barely) in the right direction. That is even more embarrassing when one considers that the Kessel-Pavelski-JvR line had some of the tougher zone starts among American forwards.
There's no doubt that a lot of hockey commentators had things to say following the Canada-USA game. While it's tough to generalize the storylines when there are so many people speaking, I noticed a few recurring ones that I'd like to address. I think these narratives are summed up nicely in this article from Ryan Lambert (similar thoughts were expressed in this article from Fear the Fin). Before I dive into this, I want to be clear: Lambert and Fear the Fin are solid hockey writers who I respect. I only reference their articles to show that I'm not responding to a straw man. With that being said, Lambert's criticisms of Team USA follow:
1. Let people who are actually good at picking teams pick teams
Ryan himself does a decent job of refuting this. While Poile hasn't been the best GM in the NHL over the last decade, he's done a remarkably good job with a tight budget. On the flip side, Yzerman's Lightning have missed the playoffs the last two years even though they've had superstar talent. Indeed, the only playoff success the team has had with Yzerman at the helm was due to facing a Penguins team without Crosby and Malkin and a goalie in Dwayne Roloson who was putting up a .950 sv% going into the eastern conference finals. That's hardly a ringing endorsement.
Moreover, the US panel had Stan Bowman, Ray Shero, and Dean Lombardi on the executive committee, architects of four of the last five Cup winning teams. On top of this, Ray Shero is the incumbent winner of the GM of the Year Award. While these men didn't have the final say, it seemed clear that they were key participants in the process and agreed on nearly all of the pieces of the US roster (as evidenced by the ESPN article which followed the selection process).
To sum it up, the US committee had some of hockey's best minds on the selection panel, and there is zero evidence that Shero, Lombardi and Bowman wanted one thing while Poile wanted something else. I'm very skeptical that changing who had the final say would have had a meaningful impact.
2. Bring your best players
Like most, Lambert was critical of the decision to leave Bobby Ryan and Keith Yandle off the roster (more on Keith Yandle below), and he took a number of shots at Brooks Orpik. The complaints about Orpik were justified because he was bad in terms of both scoring chances and possession during the Canada-USA game. But the misleading thing about the "blame Orpik only" meme is that it paints a picture where he's the only player worth criticizing and replacing. Nothing could be further from the truth.
If we're criticizing d-men, Fowler and Shattenkirk deserve a ton of blame. Both were negative in scoring chances and hovering close to 40% in Corsi. What's terrible about that is that the coaching staff did everything possible to shelter these two: they got easy competition and had zone starts between 65-70%. When your team as a whole is getting out-possessed, but the coaches find a way to start you in the offensive zone 70% of the time, you really, really need to find a way to at least draw even in possession. But all of their offensive, puck-moving skills amounted to little in this game as they spent more time in their own end because they couldn't get the puck up the ice.
For the record, I don't think, ex ante, it was wrong to bring Carlson, Fowler, and Shattenkirk this year since they're the future of the Team USA blue line. But the fact that they're young, offensive d-men shouldn't excuse them from justified criticism.
Moving to the forward group, the picture is even bleaker. TJ Oshie, David Backes, Ryan Callahan, Dustin Brown, Max Pacioretty, Paul Stastny, and Zach Parise were all dreadful. And that's not an exaggeration. Parise had a 40% Corsi For even though he got to start 70% of the time in the offensive zone; Callahan had the worst scoring chance differential of any American. And yet nearly everyone who peddled their individual story after the game ended never lobbed consistent and appropriately harsh criticism at these 7 or 8 forwards for their lackluster play (Ryan Lambert directed some criticism at Callahan and Brown in passing). Going into the tournament, no one disputed that at least six of those forwards should have been on the team, but they all turned in bad performances when it mattered most. Given that forwards have more control over possession, they should have been shouldering more blame than Orpik.
The bigger point in all of this, though, is that the US simply wasn't as talented as the Canadians, and even if they brought Bobby Ryan, Keith Yandle, etc., they'd still be undermanned. The easiest way to see this is to simply ask who on the U.S. roster could have made Canada's 2014 team (the idea for this thought experiment came from the always fantastic get to our game blog). On defense, I think only Suter makes Team Canada. The only other US d-man worth considering would be McDonagh, but Vlasic does everything he does three times as well. As for the forwards, Kane and Kessel make the team, and...that's it. I don't see any of the American centers overtaking Crosby, Getzlaf, Toews, Duchene, or Bergeron, as Kesler and Pavelski are good but not in the same league. I also can't see any other American wingers making the Canadian team; Parise might be the closest, but I'm not convinced he's better than Kunitz, Bergeron, Perry, Benn, Carter, Marleau, Nash or Sharp.
And even if we give the critics their hypothetical roster which includes Bobby Ryan, Keith Yandle, and Kyle Okposo, the calculus stays the same: none of those guys would have made Team Canada either. So at the end of the day, Team USA fielded a roster that, at best, only had 4 of 18 skaters who would have made Team Canada. The fact that the Americans lost really shouldn't have been a surprise. It's probably best to view this game not as anyone's "fault" but as the predictable outcome of a tilted matchup.
Lambert's last criticism took aim at the supposed 2-2-1 Team USA ran during the Canada game:
3. Don’t play passive
The basis for this narrative is in the post-game comments from Zach Parise. In addition to Lambert, Pittsburgh sportswriter Dejan Kovacevic made similar comments when he said that Team USA was "declawed by their own coach. They were deployed in a 1-2-2 trap that had a group of dedicated, career-long forecheckers skating backwards." I'm not sure how you can blame a coach for a 1-0 loss when you're starting with such a disparity in talent (see above), but many have in fact done so.
In evaluating this criticism, I want to approach it first from an intuitive standpoint. I think that it's beyond dispute that Canada was a faster team with more skill. When a coach is facing a team with those characteristics, and when he doesn't have similar talent in terms of speed or skill, it's not unreasonable to devise a plan that seeks to neutralize the opposing team's speed/skill by clogging up the neutral zone and making it difficult to maintain possession. That can't happen if, hypothetically, the team has two or three forecheckers below the blue line pressing the puck carrier. Indeed, many NHL teams use an aggressive forecheck, but they don't force turnovers every time they press. In fact, the success rate is below 40% most of the time. And against an incredible defense group like Team Canada's, it's even less likely that an aggressive forecheck leads to bad passes out of the d-zone.
So when you press but don't force turnovers, what are you left with? Guys like Crosby, Toews, Duchene, Sharp, Bergeron, Benn, etc. have metric tons of open ice in the neutral zone to speed through and carry the puck in cleanly to create scoring chances. Keeping two forwards back clogs the neutral zone and forces Canada to dump and chase more than they'd like to.
But that's all in theory. To see if this passive approach came through in reality, I took snapshots, spaced evenly throughout the game, of the times that Team USA had more than one forechecker pressing the puck carrier. I include a lot of photos below, not because I want to beat a dead horse, but because I don't want to be accused of cherry-picking. One other thing to note: it's not common during a game that teams get to set up and run their preferred breakout while the defense gets to use their preferred strategy as well. Line changes, bouncing pucks, goalie rebounds, and pucks off the glass all prevent controlled, planned breakouts. The point of this is that the shifts where the defense gets to run its 2-2-1, 2-1-2, or 1-3-1 against a clean breakout happen much less than people think.
Onto the pictures (please click all images to enlarge them):
It seems pretty clear that throughout all three periods of the game, the Americans had a number of shifts where they had 2 or more forwards forechecking below the blue line. They obviously didnt run the 2-2-1 the whole time, as they picked a number of spots where they thought they would be able to force turnovers with an aggressive forecheck. This didn't always work out and often led to quality scoring chances against, but I think it disproves the idea that the US played the whole game in a passive 2-2-1 setup, or that the game plan effectively "declawed" the players.
Some might respond that Bylsma is an idiot for playing in the 2-2-1 at all. Instead, the argument might go, he should have played aggressively the whole game. So I went back to see if Babcock and the Canadians, about whom no one has anything negative to say, ever had a passive forecheck. The photos follow:
Canada plainly had a number of shifts where they clearly didn't press. But no one complained about Babcock selectively employing a 2-2-1 or a 2-3.
I think three points follow from this: (1) Team USA did not execute a completely passive game plan, as they aggressively forechecked on a significant number of shifts and ultimately used mixed strategies; (2) Team Canada did not always utilize an aggressive forecheck; and (3) "systems" probably didn't have a meaningful impact on the game, as the talent differential between the two teams was so wide. Therefore, talking about Bylsma's passive approach can be charitably viewed as a red herring, or at worst, a false narrative.
The final, big story coming out of the USA-Canada game was Bylsma's "issue" with match-ups, specifically, his alleged inability to get the Kessel line away from Jonathan Toews. I had noticed before this game that a lot of people on the internet questioned Bylsma's ability to get favorable matchups for his players or to manage his bench well, which led me to write this piece debunking this myth. Pierre Lebrun, however, made matchups an issue in his post-game reactions:
Dejan Kovacevic also said that Kessel was "utterly smother[ed]" by Toews. It's easy to figure out if this claim has any merit. I simply re-watched the game and counted how many times Kessel was up against Toews and compared that to how often he was matched up against other lines. I also added one wrinkle to this mix: I wanted to see how often Crosby was matched up against David Backes. The background on this wrinkle is easy to understand. Backes has always matched up against Crosby when the Penguins play the Blues, and he tends to consistently frustrate Sid at even strength. Here's the shift chart comparison for the Penguins game in St. Louis earlier this year showing that Backes was covering Crosby.
This year, Crosby has been averaging 1.31 points per game and has a Corsi For of 52.8%. In the two games against St. Louis, Crosby had zero points and a Corsi For of 43.6%. This was not lost on the coaches heading into the Olympics:
Dan Bylsma said yesterday that he will try to have Backes on Crosby all game....with last change.— TSN 1200 (@TSN1200) February 21, 2014
Bylsma says he wants Backes vs Crosby. Babcock says he's not worried about match game, but Weber vs Kessel seems smart 2me. last change US— Ryan Rishaug (@TSNRyanRishaug) February 21, 2014
Below is the chart containing the match-up comparisons for Kessel and Crosby. Two quick notes. First, the bar for "Against Favored Matchup" means that during their shifts, Kessel was against Toews and Crosby was against Backes. "Away From Favored Matchup" means Crosby was away from Backes and Kessel was away from Toews. Second, if Crosby or Kessel had one shift where they saw 15 seconds or more of Backes/Toews and 15 seconds or more of someone else, I counted that as one shift against favored matchup and one shift away from said matchup.
I double-checked these numbers, and I think the results are damning for people like LeBrun and Kovacevic who were critical of how Bylsma managed his bench. This data brings three points to mind. First, the narrative was false. Bylsma did in fact get Kessel away from Toews: nearly half of his shifts were spent skating against one of Canada's other three forward lines. I can imagine that early on in the game, LeBrun noticed Kessel out against Toews and thought that was strange. Then confirmation bias took over and he noticed every other time Kessel was matched against Toews without noticing the nearly equal number of times that Kessel was not matched against Toews. This only underscores the importance of going back and counting shifts.
Second, getting Kessel away from Toews makes no sense because that was the only matchup that was working out positively for the Americans. If we go back up to the charts at the beginning of this article, we can see that Pavelski, JvR, and Kessel were the only forwards from Team USA with a positive scoring chance differential. Moreover, Kessel was the only player to have a Corsi % north of 50%. Conversely, Toews and Marleau were among Canada's worst players in scoring chance differential, and all of Toews-Carter-Marleau were towards the bottom in possession. Because Kessel was able to tilt the ice in his favor while out there against Toews, wouldn't Bylsma be a bad coach not to keep getting that matchup? Even if you're inclined to answer that question "no," Bylsma still got Kessel away from Toews for nearly half of his even strength shifts. I don't think we can ask for more. (To those who might say Kessel's positive numbers came when he was away from Toews, that isn't true. If it were, Toews would have better stats since Kessel is theoretically doing his damage away from Toews).
Third, this "matchup" criticism illustrates what is becoming a disturbing double standard that Bylsma is subject to. Notice how folks in the media flipped out when Kessel was matched against Toews (even though he was winning that matchup) but said nothing at all when Babcock did not try to get Crosby away from Backes despite Backes's history against him. Indeed, Babcock said he didn't care about matchups at all. Yet Bylsma was skewered by LeBrun and others for not developing his entire game plan around who Mike Babcock would deploy in any given situation. Some might say keeping Crosby away from Backes wasn't a necessity since Crosby was dominant during the semifinal game. That is true, but Kessel also "won" his matchup. The margin of victory for Phil was much smaller though since he isn't as good as Crosby, and because he didn't have the high-quality teammates Crosby got to skate with. But suffice it to say, if the media isn't going to subject Babcock to criticism for ignoring matchups, it should do the same for Bylsma.
I am devoting the final segment of this article to Keith Yandle. Many people wanted him to be on Team USA, and the guys over at Fear the Fin lamented his absence and referred to him as an "elite puck-mover." I wanted to look at Yandle's possession stats to see if he was truly as dominant as the commentary surrounding him suggested. Below is a graph comparing his zone start % with his Corsi For % (all stats from Extra Skater):
It's superficially good that Yandle is above 50% Corsi all three years, but he's getting really sheltered minutes. I actually think it's a bit of a red flag that his zone start % is so much higher than his Corsi For % in each of the last three years. Yandle isn't bad by any means, but he's not driving play in the right direction.
The picture also doesn't get any rosier when we consider his quality of competition. Yandle's ranking among Phoenix defensemen in quality of competition for 2011-2012, 2012-2013, and 2013-2014 is 6th/7, 7th/7, and 4th/6, respectively. Dave Tippet is clearly going out of his way to give Yandle the easiest of easy minutes, but his Corsi For % has never been higher than 52.5%. Given this, I can't imagine a scenario where Yandle is an "elite" puck-moving d-man. The best case scenario for him is slightly above average based on these numbers. Would he have been better than Orpik? Probably. But he's not elite, and I don't think he would have altered the final outcome of the Canada-USA game since Fowler and Shattenkirk were eating all of the sheltered minutes on D.
* * *
Though this has come roughly six weeks after the Olympics ended, I hope this piece can be a reference point going forward for those who want to clarify the record on what exactly happened in Sochi. All I've done is re-create the same basic data points that we use to evaluate NHL teams after every game. Just because there weren't advanced stat reports published after each Olympic game doesn't mean we have to rely on our memory. When something as important as the Olympics is concerned, it's essential that we make sure we're making arguments supported by data.