The Numbers Game: Loui Loui. Oh No?

A third of the way through the season (holy crap, already?), media outlets around Boston both conventional and newfangled have begun to take stock of one of their newest playthings. Though halted for few games at the hands of human shit-stain John Scott, Eriksson's performance as a Bruins is beginning to come into focus. Still on a 50-ish point pace, reactions to Eriksson's start have been positive if measured, many praising the little things he does while admitting some disappointment in his production. Meanwhile, one blogger - who may or may not have previously suggested that having Greg Campbell is preferable to Alex Ovechkin - proposed that Loui Eriksson is doing more harm than good.

That is, of course, a preposterous claim for a forward leading his team in CorsiRel - when not accounting for zone starts that is - and in even strength production per 60. (Reilly Smith, it should be noted, is 2nd. Encouraging, but slightly faint praise with no Bruin including these two above 70th in the league)

However, there's a kernel of truth to the the suggestion that Eriksson could be partially behind both his linemates producing at their respective career lows. Lets take a closer look.

-----------------------------

No, we're not here to talk about "the trade." This isn't about the relative merits of the two key players per side or the myriad moving parts involved. It's not about pining for what's lost or gloating about gains. It's not about the scoreboard between the NHL-active parts of the trade package (it's just .72/ppg to .63/ppg in the Stars favor, for those keeping tally). The following is simply reporting findings surrounding a second line not clicking at its optimal rate.

First, lets address the symptoms. Overall per-game output for Bergeron is at .48/gm (down from a career .74). He's down pretty markedly and scarcely at a 40 point pace. That's rather alarming, but against this backdrop, he's getting a tenth of a shot more per game than usual and is presently firing a percent above his career sh%. At even strength though, his rate stats sit at their lowest point since his post-concussion season in 08/09.

Marchand, meanwhile, looks similar. .44 PPG down from .59. Just off his customary shots by .3 shots/game. Lowest ES P/60 of his career. Even so, it's worth mentioning that for all the scrutiny of his game early this season, he's out-producing Iginla, Lucic and Bergeron at 5v5. Still pulled from the powerplay after his demotion earlier this season, one might think that the 1.4 minutes/60 of man advantage time he lost has made the difference - but that difference would only pad his totals by 2 points at this stage of the season.

Both Marchand and Bergeron have maintained their "very good" and "friggin excellent" respective CorsiFor%s. On-ice shooting percentage for both is well within the norm.

So what's the deal?

With all things pretty much status quo for these two players, we then consider the right wing slot. Eriksson, likewise, is proving a possession beast, as referenced above. His shooting percentages, both personal and on-ice, are right where they should be. His shot count is right around his norm. And here we start to see some difference this year:

All things otherwise equal on this line, Eriksson gets off nearly one and a half fewer shots per game than his predecessor (or 6.8/60 to 11.8/60). Doesn't sound like a big deal really - until you extrapolate that over 82 games at 10% shooting that amounts to a difference of 12 goals-for, individually. Which of course trickles down to assists for frequent lineys. Drilling down a bit further, what of his on-ice CorsiFor? Why, it's around 2 events/60 lower, but should this be cause for concern? Zoom in a touch more, lets look at his individual CorsiFor (hereafter referred to as iCorsi as represented on HockeyAnalysis.com) - simply, the shot attempts he himself takes.

-----------------------------

And there's the major difference:

Per/60, Eriksson generates more than a third less than Seguin did from his own stick. I don't want to overstate this, but for a frame of reference, that's essentially the equivalent of an entire extra player on the ice with Seguin on the line in lieu of Loui. As with assists on goals, this trickles down to fewer second attempts for Bergeron and Marchand.

However, as we've surmised, Eriksson has had a positive influence on the defense. Turning back to overall On-ice Corsi, we see the line's rating below:

As you can see, Corsi against at 5v5 has dropped by around 3 events/60, so there's certainly evidence that Eriksson has helped to improve the defense when he's on the ice. On the other hand, the line is generating around 5-7 fewer events For. Applying the league's mean 4.2% Corsi shooting percentage - simply the conversion rate on shot attempts rather than shots - and projecting for 82 games, we've lost around 5 even strength goals between the three forwards over the course of a season and gained scarcely 2 prevented on the defensive end.

In case you're wondering as I was whether all this above is just the noise of  standard year-to-year fluctuation, i confirmed the repeatability of 5v5 iCorsi for forwards, which through regression analysis comparing annual year over year league-wide figures yielded a coefficient of .57 - indicating that the generation of shot attempts is an entirely repeatable skill largely controlled by individual players, as you might expect from a component of highly-repeatable On-ice Corsi. Corsi Sh% meanwhile held a year over year r^2 value of .08 - not very correlated, so subject to strong mean regression.

Knowing the above, we understand 1) that we can expect Corsi Shooting % to regress just as reliably as conventional shooting % and 2) that players will be fairly consistent in terms of their individual shot attempts. Past research has shown CF and CA (especially) to be highly repeatable as well.

Quite simply then, we're not likely seeing a fluke. The line's lost more in offense than its gained in defense. The purported tradeoff hasn't exactly panned out as intended - though I'd stop well short of saying the new configuration has done "harm."

It's still pretty damn good.

----------

As I've said at the time of the deal, the margin between the 2nd line with Seguin and with Eriksson would be a fine one for the time being. And with the differentials above between the two iterations, this appears to be borne out. Eriksson's proven to be a highly valuable, play driving, offensively productive top six forward as advertised, albeit one who doesn't markedly elevate production for those around him.