About a month ago, I spent a few weeks toying around with a project that attempted to identify firmly, objectively, and catagorically the worst forwards in the NHL. It didn’t work out, mostly because, while I am an ardent supporter of the fancystats revolution, there are still a lot of issues to be resolved involving how one should weight various pieces of data against each other, particularly where a broad and somewhat subjective term like ‘worst’ is involved. Make up a big chart with all the columns, your Corsis and Fenwicks and zonestarts and qualcomps and this-that-and-the-others, and you can get a pretty decent idea in your own mind of who the worst players are, but proving certainly that this one is worse than that one is a difficult thing.
Of course, most of the guys who came out in the bottom are the ones you’d expect, goons and AHLers forced up by accident and injury. If the list was only those names, perhaps I’d have felt more comfortable putting it up and letting the comments fall where they may, for who on this fair internet is really going to argue with a straight face that Cam Janssen is a good NHLer? But there was also an oddity in my final list that I couldn’t quite figure out: it was packed full of Oilers.
On a list of the ten worst players in the NHL, I had four guys out of Edmonton: Anton Lander, Ben Eager, Darcy Hordichuk, and Lennart Petrell. These guys all had miserable stats in almost every area. As in, bottom 10% of the NHL bad. Worse than Krys Barch bad. Almost as bad as Cam Janssen bad. You want see table? Here, see table. Is bad.
Now, you may believe there are problems with fancystats, but we should be able to agree that they measure real things, important facts about how hockey players are used and what they do. You can debate whether this number is more important than that one, or this marginal difference is important or not, but what should be beyond debate is that, given certain inputs (sufficient ice time, easy opposition, shifts started in the offensive zone) a decent player should be able to generate some sort of positive outputs (more shots/shot attempts for than against, more penalties drawn than taken, more Ozone starts for the line that follows). This isn’t arcane numeromancy. That’s basic hockey: good players can do good things with good minutes. Doing bad things with good minutes is more or less the definition of being bad at hockey.
Nevertheless, when you get four players on one team all very nearly bottoming out the NHL in badness, it suggests that the problem is something more than just the individuals. Granted, it is possible that Steve Tambellini is literally the worst judge of fourth line talent in modern hockey, but it seems more likely that what we’re seeing here is some kind of team effect. Hockey players’ relationships with their linemates are an extraordinarily complex lattice of influences that goes far beyond individual skills and decisions. In both action and inaction, one player affects the options available to the others he shares the ice with, and this goes both ways: good players make their linemates better, bad players make their linemates worse. And, just as sometimes a good center makes his winger look better than he actually is, sometimes a miserable winger makes his center look crappier than he could be.
But how do you know which way the influences go? If you are Tambellini, or let’s say the better GM who gets hired when Tambellini finally wears out his welcome, how do you disentangle this web of suckery that has formed on your fourth line?
Whoa whoa whoa, you might say, why do I want to disentangle this mess at all? They’re the fourth line. Fourth lines are always shitty. That’s what they are. Who gives a fuck if my five-minute-a-night-guys are worse than other five-minute-a-night guys? I’ve got Taylor Goddamn Hall to fix everything, right?
Well, for one thing, if you’re the Oilers you’re playing most of these guys (except for Hordichuk) between eight and nine ES minutes per night, not five. But even if you weren’t, there is an under-researched phenomenon in of inter-line interdependence in hockey. Just as a better-than-normal first line can create echoes throughout the roster that improve the whole team, a worse-than-normal fourth line reverberates through the depth chart in ways that hurt it. You want a fourth line that can, at absolute minimum, hold its own against other 3rd/4th lines. If your shitty players are taking offensive zone starts against easy opposition and getting killed, it means more shots against, more chances against, and ultimately more goals against. Moreover, though, soft minutes being chewed up by the fourth line are soft minutes that aren’t available to your first and second line players. If I’m the Oilers and I’m planning to spend next year trying to prop up Jordan Eberle’s image as a 70-point-player, I want him, not Anton Lander, on the ice against the shittiest opponents. A half-decent fourth line gives your coach options. A terrible fourth line deprives him of them.
So how do you figure out which of these guys are causing this clusterfuck and which are merely trapped on the bottom of the pile, unable to wiggle free? Does it help to consider the WOWYs (With or Without You, a measure of how players did playing together and separately)? Does it help to consider the reasoned opinions of dedicated and knowledgable observers?
Anton Lander, as we can see, got well and truly murdered pretty much across the board. If he wasn’t the worst on the team in a given stats category, he was the second worst. Clear-cut case of taking good inputs and making them into terrible, terrible outputs. Dump him, right? But wait, wait- Lander was a rookie last year, and the smart men seem to think the major problem isn’t a lack of ability but rather premature promotion. It’s just not reasonable to expect a first year guy to drive the bus, particularly not such a horribly broken down wreck of a bus. And hey, although as we shall see he was speculatively blamed for Petrell’s problems, he didn’t make everyone worse- Hordichuk was actually slightly better with him than without him.
Ben Eager is exactly the sort of guy that statsvolk usually want to see excised from a team- big, fighty, hands suited to punchin’ not passin’. And dude is a bad player, no doubt, but his stats are actually not as miserable as Lander’s. He took some slightly more difficult minutes and got some slightly better results. He’ll always be in the ‘fighter’ category and you can certainly argue (as I do) that teams shouldn’t have a designated fighter, but fact is plenty of ‘em do and manage to succeed. The Devils manage to strategically carry two terrible players (Janssen and Boulton, both worse than pretty much anything on the Oilers) without the contagion spreading to the rest of the roster. Eager might be a problem, but he’s a common, ordinary, routine kind of NHL team problem that half the GMs in the League wouldn’t even consider such. That’s not the same as being the problem.
Darcy Hordichuk looks like Eager only more so, playing the fewest games and the fewest minutes against the easiest opponents, but in several categories he got the least terrible results. Moreover, all three of the other guys were better with him than without him. Could a guy who doesn’t play very much and seems to make bad teammates marginally better when he does actually crack the roster, somehow, secretly, be having the opposite effect? I dunno. As much as I have been known to blame the combative types for everything that’s wrong with hockey, I can’t figure out a case where Hordichuk is the linchpin of awfulness in Edmonton.
So maybe it does come down to Lennart Petrell. Petrell was really, really, really, really bad this season. Bad beyond bad. Look at that chart. See that -22.48 in the CorsiOn column? That is the worst number for a forward in the NHL. The Oilers are getting out shot-attempted by more than twenty freaking shot attempts for every sixty minutes this guy plays. All the other bad players are less bad when they’re not playing with him. There’s almost no way to slice the data that makes Petrell look good, except that during a brief stretch when he played without Lander and saw some time with top-six guys, he miraculously became better for a while. But does that proof of Lander’s power to drag Petrell down, or other players’ power to drag him up? Regardless the evidence against him, though, the men-who-watch seem to like dear Lennart quite a bit. It’s an odd situation where the goons are actually getting slightly better results than the guys with the better skills, and poses the difficult question of which is more worth having faith in: results without potential, or potential without results?
For mine own part, I’d be tempted to just blow up the bottom of this roster and experiment with entirely new pieces. Scorched earth. Demote ‘em all and let OKC sort it out. But, between you and me, if we are going to be entirely honest with each other, we all know that blowing shit up is the lazy fan’s solution to complex problems, and most of the time it doesn’t work anyway. So what is the mature, professional, adult, general managerly way to address this problem? How would you solve a problem like Petrell, or Lander, or Hordichuk, or Eager?
Stats courtesy of behindthenet.ca. I apologize for probably doing it wrong.