Crosby and Malkin Together Again

FOREWARD:  Below is the article that ran in the Toronto Star this week. But I wanted to add just a little more discussion that wasn't included due to space limitations, but which I think it important to properly analyze the issue raised.

First, although the article mentioned this, it's worth highlighting that Crosby and Malkin played together 5v5 for only 201 minutes over the past 3 regular seasons.  That's not a lot (though it's also not a little -- that's about what regular linemates would play over about 13-15 games). I'd bet that with a larger sample the offensive productivity of Crosby/Malkin/Kunitz would turn positive. Although to be fair, since be baseline is Crosby/Kunitz, that's a high standard to surpass.

Second, because of limitations in the publicly available data, it wasn't possible to precisely measure the impact of linemate changes on each individual player's productivity or overall performance. So I used goals scored as a proxy for offensive output. The proxy isn't perfect -- e.g., it doesn't account for additional goals scored by defensemen -- but it's a good start.

Third, despite the Star's headline (the Star writes the headlines), and despite the numbers and even conclusion of the article, I wouldn't go so far as to say that it's never correct to play Crosby with Malkin together for a regular shift. But the conditions would have to be just right. Specifically, I think that playing Crosby/Malkin/Kunitz together, with the attendant decrease in productivity in other lines might make sense if the following conditions are met:

(1) The opposition lacks a shut-down line/D-man;

(2) The opposition's line combos are relatively even in defensive ability; and

(3) The opposition doesn't have one big offensive line that might take advantage of the reduced defense of the big-three on a line together.

My thinking is that if the big trio can overwhelm any defense thrown out against it, and separating Crosby and Malkin would only dilute the offense against a team that's relatively uniform in defensive ability, then that could be successful.  As it happens, that sure sounds like the Rangers. Note that this also means that if Bylsma follows the same analysis his decision to move Malkin to Crosby's line might well not have been primarily for the purpose of getting Crosby back on track, as my article implies.

I'm interested to see whether the Pens stick with that strategy if they make it through to the next round, in particular against the Bruins, who almost certainly would be matching Chara and Bergeron against them

With that, here's the article:


Until Monday night it had been 13 straight playoff games since Sidney Crosby had lit the lamp.

Crosby’s drought gave rise to an interesting decision for Penguins coach Dan Bylsma: Should he stay the course and keep together his regular top 2 lines of (1) Crosby-Chris Kunitz- Lee Stempniak, and (1a) Evgeni Malkin-James Neal-Jussi Jokinen; or should he shake things up and put his three big guns together in one uber line of Crosby-Malkin-Kunitz?


For excitement the uber line is definitely the way to go. But does it make sense if you’re trying to win playoff games, in particular when no player on the roster outside these 6 scored more than 13 goals this season, so if things go wrong, there’s no cavalry coming to the rescue?


The contours of the analysis are relatively straightforward:


  1. Determine whether any expected increase in offensive production by Crosby/Malkin/Kunitz is greater than the loss of production suffered by their usual linemates, Stempniak, Neal, and Jokinen;


  1. Factor in any expected drop-off in defensive performance; and


  1. Estimate how much it’s worth to get Crosby off the schneid. Will it propel Crosby to a tidal wave of goals while allowing you to move Malkin back with his regular linemates for some secondary scoring?


To determine whether moving Malkin should be expected to result in increased offensive production, I looked at past performance. Malkin, Crosby, and Kunitz have played together for around 200 minutes in 5-on-5 situations over the last three seasons. This has been great for Malkin, whose goals scored per 60 minutes (G/60) jumped from 1.15 to 2.09. But it actually resulted in a significant decrease for both Crosby (from 1.10 to 0.60) and Kunitz (from 1.41 to 0.78).


So in the aggregate, adding Malkin to the top line has -- surprisingly -- resulted in 0.19 G/60 fewer goals for the three of them, albeit in a fairly small sample.


If the idea is to get Malkin to score, then this has been a great strategy, but if it’s to help Crosby, or even increase offensive production across the three players, then not so much.


On the other side of the equation, playing without Crosby/Malkin – unsurprisingly -- results in big drops in G/60 for Stempniak (from 1.00 to 0.41), Neal (from 1.25 to 0.94), and Jokinen (from 0.84to 0.67), for an aggregate decrease of 1.07 G/60.


Overall then, by playing Malkin with Crosby and Kunitz the Penguins are looking at an expected loss of production of 1.26 (-0.19 + -1.07) goals per 60 minutes.


If anything, the actual number may be even worse for at least two reasons. First, in the past the Penguins have played Malkin with Crosby sparingly – since 2011 Crosby has played only about 9.5% of his even-strength minutes with Malkin. Presumably these minutes were played in circumstances favorable for offensive production; for example, if the Pens were behind and needed goals, or if the opposition’s shut down defenseman wasn’t on the ice. Which means that the offensive output numbers for the two as linemates are probably inflated, in particular relative to what might be expected in tougher, tighter-checking, playoff games.


Second, when Crosby and Malkin play together they tend to give up more goals against. The Penguins give up 3.28 G/60 when Crosby and Malkin play together, versus only 2.23 G/60 when Crosby plays without Malkin. We should discount those numbers somewhat, since they’re probably at least partially a product of the Penguins using the two together only when they desperately need goals and so are in a chance-taking mode. But still, that’s a pretty big drop.


Which leaves the final factor I mentioned above: what’s it worth to get Crosby on the scoresheet? This too is something that advanced stats can help answer and it’s on the Department of Hockey Analytics’ long to-do list. We expect to be able to determine what effect, if any, “momentum” or “confidence” or “streaks” have on a player’s production.


But we haven’t performed that analysis yet. So for now, suffice it to say that it seems highly unlikely that whatever confidence the best player in the world might gain from breaking out of his slump would be worth the 1.26G/60 – at minimum -- the Penguins are giving up by playing him with Malkin (not to mention the potential drop in defensive performance). I doubt Bylsma, or anyone watching the last two Olympic gold medal finals, believes Crosby’s psyche is so fragile that he needs such a confidence boost.


I’m all for betting big when the situation calls for it. But the risks have to be justified by the expected benefit. In this case it’s hard to see how the Penguins putting all their eggs in one basket to help get Crosby back on track was worth the risk.


No Comments Yet

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>