Taken at face value it’s a simple question: What makes a hockey defenseman valuable?
Think about that for a second.
Is it skating ability? Passing prowess? What about the quality of his shot?
How much does point total matter? More would seem to be better, but ‘defensive defensemen’ are valuable, too.
Should body checks or physicality factor into value? What about shot-blocking skills? The ability to draw penalties is always a plus and so are skills on special teams.
Now for another, much harder question:
How would you go about quantifying all those things plus the things I didn’t mention into a single, easy to understand number for each individual player while adjusting for each player’s quality of competition AND quality of teammates while also factoring in hockey’s heaping portion of luck-based events that happen on every shift in every game?
I’m not sure about you, but I just had a panic attack.
Timo Seppa has not, apparently, and that’s good news as he leads a group of men tasked with the seemingly impossible–bringing statistical order to something inherently chaotic, namely the game of hockey.
And believe it or not, they’ve come up with something.
It’s called ‘GVT’ (Goals Versus Threshold), and it attempts to bring the idea of “Replacement Level” to hockey analytics, much in the same way ‘VORP’ (Value Over Replacement Player) did in baseball.
“Replacement Level”, by the way, is defined as the level of performance a team can expect when trying to replace a player at minimal cost.
From “Hockey Prospectus 2011”, somewhat paraphrased:
“Replacement Level’, in other words, describes the value of players that get called up when a player is injured. In the NHL, 13th forwards or seventh defensemen are considered replacement level, as are good AHL players. A replacement level team would get outscored by 1.5 goals per game on average, or 123 goals over the course of a season.
Now, GVT in action:
Sidney Crosby posted a 30.0 GVT in 2009-2010, his most recent full season, putting him among the league leaders. So he was worth 30 goals over and above what a “replacement level” player would contribute if were brought in to take Crosby’s place on the roster, most likely from the minor leagues or via the dregs of free agency.
You’re probably wondering why Crosby was only worth 30.0 GVT in ’09-’10 when he scored 51 goals that season. Me too.
More explanation from Hockey Prospectus’ Tom Awad, creator of GVT:
Offensive GVT is extremely straightforward: it measures a player’s contribution to scoring goals. Offensive GVT is based on goals and assists scored above what a replacement-level player would have done with the same ice time. That last part is key; players who play more games or get more ice time, especially power play ice time, are expected to score more, and should not get credit for inflated scoring totals. GVT also assumes that goals and assists are not of equal value, and assigns a weight of 1.5 to goals relative to assists.
This produces offensive GVT numbers that don’t exactly match point totals. Sidney Crosby, despite missing half of last season and consequently finishing outside the top 30 scorers, finished ninth in offensive GVT, which indicates how much more impressive and valuable his 66-point performance in 41 games was as compared to, for example, Mike Ribeiro’s 71 points in 82 games. Crosby’s offensive GVT was 15.0 while Ribeiro’s was 9.3.
Notice the qualifier, “Offensive” in front of GVT in the excerpt. This is where the burgeoning metric runs into some clarity issues, as it’s broken down to Offensive, Defensive, Goaltending and Shootout GVT, with Defensive GVT being the least credible at this point.
“It’s the stickiest one,” Seppa said. “We’re working on it. The point of it is to measure a defender’s ability to limit shots, which is key to preventing goals.”
It seems our initial questions regarding defensemen may be tougher than we realized, even for the experts.
Teething issues with GVT aside, Seppa’s right on that last count–perhaps the most important single finding in hockey performance analysis has been grasping the value of shooting the puck.
But that’s another issue for another time.
For now, a closing note from Timo:
“It’s hard to wrap your head around all the moving parts of this sport, but that’s both the challenge and the promise of sabermetrics in hockey. We’re making progress. It’s not where it needs to be, but we’re getting there.”
