Friday, September 23, 2011

Yeah...sez you! No...sez YOU!!

Yesterday, there was a brief online argument on the divide between those who “see” the game and those who “measure” the game.  As often happens in internet arguments, the fact that the argument was brief did not detract from its intensity.  The genesis of the argument was an article by Fox Sports columnist Jason Whitlock.

The opening in that column isn’t subtle: “I won’t be going to see 'Moneyball.' The movie celebrates the plague ruining sports: sabermetrics.'”  “Ruining” the game?  If Whitlock intended to provoke, mission accomplished.  But actually, this is just the latest installment of the “stats are for losers” versus “your eyes are lying to you” argument that you are likely to find on any message board, online forum, or sport talk call-in show concerning any sport you care to name. 

But the problems with the “stats versus eyes” argument crsytalized in my mind listening to this on…

It wasn't the predictable opinion that the Pittsburgh Penguins would be the Eastern Conference representative in the Stanley Cup finals that grabbed my attention (and yes, being a Caps fan I just rolled my eyes), it was these comments from Dan Rosen and E.J. Hradek concerning Evgeni Malkin, who is returning from a knee injury that ended his 2010-2011 season after 43 games…

Rosen: “Malkin is back – healthier, stronger, hungrier – he could be in line to have a huge year for [the Penguins].”

Hradek: “Malkin, I think, is ready to have a big year…”

A huge year?  A big year?  And on what information would one naturally reach this conclusion?  Let’s start drilling…

Malkin had a “big” year, a “huge” year in 2008-2009 when the Penguins won the Stanley Cup.  He played in all 82 regular season games in which he was 35-78-113, leading the league in scoring and finishing second in the Hart Trophy voting as league’s most valuable player (some other Russian won it).  He went on to dominate in the post-season, leading all players in scoring (14-22-36 in 24 games) and skating off with the Conn Smythe Trophy as the most valuable player of the post season.

But since then?  In 110 regular season games Malkin was 43-71-114, minus-10.  That is a 32-53-85, minus-7 pace per 82 games.  And that is after having missed 15 games in 2009-2010 to injury and 39 games to injury last season.

At this point we’re thinking what Rosen and Hradek “see” is the 2008-2009 Malkin – who at age 22 put up what is to date his career year – not the Malkin of the past two seasons who was not that player.

Drill a little further into the numbers…

Since that whopper of a 2008-2009 regular season, Malkin’s numbers have deteriorated.  Not to the point where he might be considered “just another player,” but certainly enough to be noticeable.  For instance (numbers from…

The first thing to notice is the games – from 82 to 67 to 43.  Even with the games, there is the declining ice time per 60 minutes at 5-on-5.  His goals per 60 minutes at 5-on-5 do not show a clear pattern, but those assists – primarily primary assists, the bread and butter for a center – drop significantly, which in turn is reflected in his points per 60 minutes dropping substantially.  He might be coming back, as Rosen put it, “healthier, stronger, hungrier,” and that might be relative to the 2010-2011 season or even the 2009-2010 season, but to the 2008-2009 level of health, strength, and hunger? 

Drill further…

The Corsi numbers tell an odd story, and perhaps a concern to a Pens fan.  Corsi is, to simplify things, a function of shots directed toward the net, for and against.  Sort of a plus-minus on steroids that attempts to account for the effect of teammates on production.  Higher Corsi values suggest being a part of more adept puck possession and of an ability to exert more pressure on opponents than they are able to exert on your team. 

Back to Malkin.  In 2008-2009 his “relative” Corsi value – that which accounts for his Corsi value when on the ice versus that when off the ice – was relatively poor.  Puck possession appeared to be better when he was not on the ice.  At 5-on-5 he was 11th among Penguin forwards playing in at least 40 games.  That measure improved in each of the next two seasons to where he led the team in relative Corsi at 5-on-5 in the 2010-2011 season.  But despite being a part of this shot domination, and by extension puck possession, Malkin was an inefficient scorer – 2.17 points per 60 minutes at 5-on-5.  That works out to 22 even strength points for his 43 games played…before his knee injury that ended his season (although to be fair, he did sit out ten games over two separate stretches before his season ender with other knee injuries). 

We are not seeing how, based on Malkin’s revealed performance over the two seasons following his career year, one could conclude that it is even probable that he finishes the 2011-2012 season with a “huge” year.  If I’m a betting person, I’d be thinking he will end up at being about a point-a-game player over 70-75 games.  Better, certainly, than the 2009-2010 or 2010-2011 seasons (because he might be healthier or stronger), but not up to the “huge” year he had in 2008-2009.  Since the level of performance reflected in his numbers, even when healthy, doesn’t argue for it.

We could go on drilling, and there are folks out there who could do it far more effectively than we could, but the point is not to cast aspersions on Malkin's ability to produce.  It is that there are two sides in this larger argument competing against one another when perhaps they do not have to.  In this instance, perhaps a non-representative one since it involves one player on one team over a three-year period, the eyes see one thing, while the numbers reflect another.  But do they have to be “competing” arguments?  Can they be “complementary” arguments or approaches that inform one another?

Both sides have their strengths and their weaknesses.  The folks of the old school who “see” the game lack the rigorous structure and method that an attention to data brings, or they are too uninformed about data and method to appreciate its potential contribution.  And the numbers folks with their “fancystats” sometimes lack the texture that watching the game entails and conflate a player’s “production” (does he “put up numbers?”) with a team’s “performance” (do his numbers contribute to or result in winning?).

Statistics are the revealed production of the game’s participants distilled into numbers that can be manipulated and molded into models of performance.  The observers of hockey have the stored memory of hundreds – thousands – of observations of players in different contexts, different situations, with different teammates or cities, and under different emotional or other intangible circumstances.  If those who “see” the game cannot inform the “stats geeks” as to why their observations are more important than the rigor a numbers approach brings to the game; if those with a command of and a fondness for numbers cannot teach those who are empirically challenged why methodology matters; then each side is going to be talking past one another, sneering at the other’s approach to the game. 

And we are going to see a lot more insufferable columns such as Whitlock’s.


Anonymous said...

Regarding Malkin's Corsi Rel, looking at competition

2007: middling on Pens
2008: highest on Pens
2009: middling
2010: lowest

Explains part of it, at least.

exwhaler said...

As somebody who's edited statistics for a living for a decade, I'm not a fan of Corsi at all as a measure of player performance. It's an incredibly flawed statistic that ignores far too many variables.

Hooks Orpik said...

2 pre-season games, 2 goals, 3 assists for Malkin. I'd argue, that is what they saw. Malkin was the best player on the ice in two mixed-talent-level exhibition games and looks to be in excellent shape, conditioning with little traces of leg/knee problems.

I think Malkin is in a position to be better than his past two seasons, in which he had terrible line-mates in 09-10 and nagging knee/shoulder ailments in 10-11 even before the big injury. I don't think he will win the Art Ross, but I do think he ought to revert closer to the top of the heap where someone as talented as he is belongs.

The Peerless said... is probably an improvement on "plus/minus," since it attempts to incorporate the interactive aspects of what happens when five skaters are against five skaters, but as a general statistic it suffers the flaws of being just so general.