How do we judge player failure in the NHL? (Trending Topics)

Earlier this week, the Montreal Canadiens sent Alex Semin to the AHL, and then terminated his contract so that he could go to the KHL instead.

He'll play the rest of the season for Metallurg Magnitogorsk (assuming they don't get sick of him, too). That makes three pro teams for Semin in the last eight months or so, and two of them were so sick of him they just let him walk. In fact, Semin is still collecting checks from his Carolina buyout to the tune of $2.33 million, and will be until 2021.

He is such an interesting case in the old school versus new school way of evaluating hockey, perhaps the most important one.

The underlying numbers all say he's good or even excellent. And to some extent, so too do the more traditional numbers, because up until a few days ago, he was 30th among active players in points per game, at 0.8 per night. Which explains why, after he played his way out of Washington (and boy will some Caps people badmouth him at the slightest opportunity) and only earned a one-year deal with Carolina, then a five-year deal. Both were worth a $7 million AAV to Jim Rutherford, who's pretty old-school as these things tend to go.

But on that one-year show-me contract, he put up 44 points in a lockout-shortened 44 games, and even in 2013 was basically a 50-plus point-scorer if he'd stayed healthy. The production suffered the next year, even if the possession was still dominant — a score-adjusted 54.6 percent — but Carolina decided it had enough. So off he goes to Montreal, where he's scratched a ton of times, only has a single goal in 15 games, still dominates possession (56.6 percent) and is told, basically, not to bother coming back. On a $1.1 million AAV.

This seems an awful lot like the misevaluation of a player, because while you can certainly say he's been unlucky to some extent, he's also not doing particularly well for himself. In 15 games, and yes, limited minutes, he only generated two shot attempts for himself from high-danger areas. Two high-quality scoring chances from a player with his talent level, regardless of usage, just isn't acceptable. And it's therefore not going to matter very much to his employer that they continue to get more high-danger attempts per 60 minutes than their opponents when he's on the ice.

Semin got run out of three different towns in five years for his lack of results despite the good underlying numbers when he was on the ice, and to some extent you have to say the issues extend to how he behaves in the locker room. It's hard for a lot of people to believe in on-ice intangibles, but when you hear, “his teammates hated him, though,” you can see an issue. Not sure that was necessarily the case with Semin in Montreal, but there's been enough smoke over his career to speculate there was and perhaps still is fire.

However, you'd have to say another part of this relates back to expectations. You expect Alex Semin to score you a lot of goals, even if you're only paying $1.1 million for him. Obviously in Carolina the expectation is heightened by his pay grade, but players can only be what they are, regardless of salary. We learned this in the case of David Clarkson, for example. Pay Clarkson $1 million and he probably produces relatively similar results to if you paid him $10 million. Semin is no different.

In a Twitter conversation about player evaluation yesterday, TSN's Scott Cullen pointed out that Stan Bowman talked about player expectations at the Sloan Sports Analytics Conference, and noted that when Chicago asked coaches to evaluate players after every game on a 1-5 scale, they routinely rated low-skill grinders above high-skill scorers.

Put another way: If we can agree that Patrick Sharp, just to stick with a recent Chicago example, is better than Bryan Bickell, why would coaches say Bickell had a better game than Sharp more often than not? You have to think that part of it is related to the fact that it is incredibly hard to score in the NHL; last year's scoring leader barely scored a point a game, and so if a guy like Sharp doesn't produce a point every night, he's not necessarily going to look like he had a good game. But if all you ask of Bickell is to go out and throw his body around, he's going to be really successful at that, and occasionally chip in a point or two, it's because it's really easy to check people (and sometimes score a goal when you play for a team coached by Joel Quenneville).

But Bickell was sent down this year for an extended period, wasn't he? Why? Because he now has a $4 million cap hit, and he cannot justify being the checking guy alone. But in many other ways he was only having a somewhat down year by his own traditional standards, and the 94.5 PDO wasn't helping him look like a positive contributor.

That's the disconnect between Traditional Hockey People and Advanced Stats Spreadsheet-Lookers, but it's natural to look at results over process, especially when there are millions of dollars involved and guys aren't producing to expectations.

The other example of this problem to come down the pike this week was Edmonton's decision to assign Mark Fayne to the AHL. Fayne is another guy the analytics usually love, especially prior to his signing up to play in haunted defenseman graveyard Rexall Place. Even as his usage got more difficult, he has been a pretty strong defenseman in a number of ways, most of which are in the underlying numbers. But all anyone sees is a defenseman signed for this season and two more afterward at $3.625 million who has been on the ice for 68 goals against and only 39 for over 98 games in Edmonton.

It doesn't matter that he's playing some of the toughest competition Edmonton can find for him every night, a 36.4 percent goal share isn't going to be good enough. It also doesn't matter that he's probably the second- or perhaps third-best defenseman the team has (depending upon your opinion of Oscar Klefbom and Andrej Sekera).

This is also irrespective of the fact that Edmonton's goalies can only stop .899 at 5-on-5 when he's on the ice over the past two seasons. Only Klefbom's is worse (.891) over that span, but his goals-for rate is about 30 percent better, so despite a slightly worse goals-against rate and the easier usage you'd expect to see a young player given, he's seen as a potential franchise cornerstone going forward.

But the thing is, if you look at the Oilers' own-zone play this year and identify Mark Fayne as the big problem, you've missed a couple of critical steps on the checklist above him. Justin Schultz gets chance after chance — and more money — despite being worse in a lot of ways because of who he plays with. Nikita Nikitin and Eric Gryba are guys this team went out and got for some reason, and they're both flat-out awful. Nikitin is at least in the AHL (with his $4.5 million cap hit!).

Meanwhile, the team trades Jeff Petry, another guy the analytics community loved, and all of a sudden everyone in Montreal is shocked. “Wow, this guy is really, really good!” And hey, didn't that same thing happen with Tom Gilbert after the Oilers traded him to Minnesota for Nick Schultz of all people? Crazy how that happens. Over and over.

At some point maybe you have to think this is an “Edmonton isn't good at evaluating defensemen" problem, and not a “We somehow seem to get a lot of bad defenseman on terrible contracts” problem. Then again, maybe nobody's all that good at evaluating defensemen.

One of the big criticisms of Fayne this year is that he turns the puck over a lot. That can be a problem, obviously. But it ties into the larger point about playing to expectations. You see it to a lesser extent in players who are clearly valuable. No amount of All-Star appearances or Norris votes are going to convince some people that Dustin Byfuglien or P.K. Subban or even Erik Karlsson (Karlsson for Hart, by the way) are Good Defensemen in the way they understand that defensemen are good. They, too, turn the puck over a lot. And why do they turn the puck over a lot? Because they have the puck a lot, and if they turn it over occasionally, it creates the illusion that all they do is turn the puck over.

I like to think about it this way: Let's say a theoretical defenseman carries the puck out of his defensive zone 100 times, and let's say he probably turns it over 20-25 times. And while not every successful carry results in a scoring chance, that's still a 75-80 percent success rate. You would imagine that rivals just about any of the best puck carriers in the league. But because he's a defenseman carrying the puck, that means his turnovers are more likely to result in high-danger chances and, therefore, goals.

Does that mean he shouldn't carry the puck, or should play it safe? It's hard to quantify in real time, but we have to keep in mind that zone entries with a player possessing the puck result in about twice as many shot attempts within a short time as zone entries with a dump-in. The value of those 75 or 80 carries, then, is extremely high, even if the failures are probably more glaring. Which leads to people saying things like, “Karlsson isn't good defensively!” even if the rates at which Ottawa allows shot attempts, high-danger chances, and goals are all much lower when he's on the ice than when he's not. On good teams, the guys who are supposed to support players who play like him are able to do it successfully. On bad teams, they are very much not.

That's not to say Fayne is anywhere near their level, but he's a good middle-pairing defenseman, and has little in the way of a strong supporting cast on the blue line. So yes he probably does turn the puck over more than other defensemen on the team, but you have to have it to do that, and there aren't too many other Oilers defensemen who have it with any kind of regularity.

This disconnect may continue to exist for years, maybe forever, despite the Advanced Stats Revolution. To some extent it's a failure of evaluation, but to a larger one, it's human nature. And that doesn't change just because someone figured out the value of counting shot attempts.

Ryan Lambert is a Puck Daddy columnist. His email is here and his Twitter is here.

All stats via War on Ice unless otherwise stated.

MORE FROM YAHOO HOCKEY