A Million And One Questions: Of Dany Heatley, and No Pun Intended

From Wikipedia:
Jumping the shark is an idiom, first employed to describe a moment in the evolution of a television show when it begins a decline in quality that is beyond recovery.

Which brings us to this guy.

There is no doubt that Mr. Heatley has terrific career numbers; and that he's widely regarded as being a part of the upper crust of the league- amongst the elite. A potent goal scorer, a game changer, and someone who should bolster a perennially anemic and flaccid offense.

Is this really the case though?

Kent Wilson, who writes for a handful of websites, wrote a piece in Early July about the decline of Dany Heatley. Wilson used a variety of sabermetrics, which have established a solid foot hold in the hockey community (thanks to guys like Timo Seppa at Puck Prospectus,) to illustrate that Heatley is trending downward. His even strength numbers, at one point were considered elite, have slid into mediocrity. Wilson even made the point that Heatley's metrics were among the lowest on a stacked Sharks team, and comparably speaking, Martin Havlat's numbers were fantastic.

So, has Dany Heatley jumped the shark?

Full disclosure, I'm not a numbers guy- never excelled in math in school (in fact I was TERRIBLE with it,) and try to avoid them to a large extent- I even initially chose a major in College because I was told that there was very little math involved. I've come to accept the increasing usage of metrics in professional sports, namely baseball and now hockey, as an inevitability- so I try not an look at them with complete disdain. There is a place for them, in terms of widening the horizon of analysis, but I don't necessarily subscribe to things like Relative Corsi (even though I really should.) I don't think that the numbers necessarily tell the whole story, that there can be non-quantitative and non-numeric elements to the game; almost an esoteric level.

And maybe, in Dany Heatley's defense, this may be the case. Heatley had injuries to both his ankle and hand, amongst other little nagging nuisances that held his production back. Maybe the chemistry, despite how great it looks on paper, wasn't there. It could have been any number of things- elements of the game that can't be defined by metrics.Maybe the most important thing to consider here is that this is the first time Heatley's been traded; not asked to be traded, but flat-out traded. He's been savaged for being a no-show in the playoffs, and he and his 7.5 million dollar contract has been shipped here to Minnesota, for a guy that San Jose presumes can put them over the top in Martin Havlat. That should be motivation itself- to prove everyone wrong.


  1. To say that the hockey variety of Sabermetrics is worthless would be to insult worthless things everywhere.

    Corsi and relCorsi technically have their uses, but they're not really any better than the much maligned +/- stat. Sabermetrics works in baseball because everything is quantitative. I can watch a play and tell you who did something positive, and who did something negative. There's only so many plays in baseball, so I can measure every single one of them and tabulate them into a database for analysis.

    Hockey is different. The game is continuous. While you can measure things like goals, assists, time on ice, and even some relative quality of competition numbers, that's pretty much where the quantitative analysis stops. Hits and even shots aren't measured consistently (the most important quality of statistical analysis). Take a simple situation as an example: Team A has a 2 on 1 breakaway, makes the pass and gets a shot on goal. How do you measure that numerically? The defenseman played the guy with the puck, forcing a pass that resulted in a shot that the goaltender made the save on. Was that a positive play? Should he be rewarded because there was no goal, or should he be punished because he allowed a shot? He could have taken away the pass, allowing a shot from the puck carrier, but would that have scored? Was the goalie in a better position to save that potential shot? The 2 on 1 was probably his D partner's fault anyway, because he pinched when he shouldn't have. Or maybe it was a botched line change that could have resulted in a 2 on 0 but he had enough vision and hustle to get back and break it up. Or how about another play, did that player intentionally miss the net wide as a pass through traffic, or was it a missed shot that just happened to find his teammate's stick?

    Someone might claim they could measure all of these things, and establish fair, consistent quantities to each facet. Even if they manage to measure these individual plays, a 60 minute hockey game could easily have 300+ plays of this sort. We're simply not at a position where any of this data is available.

    Even the beloved Corsi and QualComp are poor indicators of performance. For Corsi, if a player is on the ice when a teammate allows a long, clean shot that the goalie sees the whole way for an easy save, that's rated as identical to a player completely giving up his man to chase the puck, allowing a pass through for an easy tap in. QualComp assumes that players are consistent. It assumes there are no players who dominate for 3 games then disappear for 6. It assumes that a guy playing with a wrist injury always plays like the average of his healthy and injured self. It ignores teams coming down with the flu. Make a mistake in the second game of a back to back on a road trip? That's averaged through your entire season, even though it doesn't reflect the quality of your play. Have a run in with a certain ref (*cough*burrows*cough*) and he calls phantom penalty after phantom penalty on you? There's so many things that aren't accounted for and just averaged out. There are spots these stats are still useful for, but they're not the uses people try to shoe horn them into.

  2. all bow down to the might and logic of Squidz.