May 11, 2012
Howe and Why
Selke Selection

by Robert Vollman

Best defensive forward

Every year, the Selke Award is given out to the "National Hockey League forward who demonstrates the most skill in the defensive component of the game," as selected by the Professional Hockey Writers Association. It's a very subjective choice for almost all of its 180 voting members, but certainly not because of the inability to add some objective criteria, something we'll demonstrate today by looking at 20 leading candidates using a dozen different defensive metrics.

Backes, Bergeron, and Datsyuk

Believing perhaps that offense is the best defense, the PHWA typically favors two-way forwards rather than the strictly defensive-minded skaters. Last year's winner Ryan Kesler, for instance, scored 41 goals and 73 points, Pavel Datsyuk averaged 30 goals and 88 points the three preceding seasons he won it, and Rod Brind'Amour averaged 29 goals and 76 points the first two years after the lockout. Even when you look way back, Kris Draper scored 40 points the year he won his Selke, Guy Carbonneau 39, John Madden 38, and even Bob Gainey averaged 37 over his four victorious seasons.

This year's finalists are no exception, with David Backes scoring 54 points, Patrice Bergeron 64, and Pavel Datsyuk 67. Here's a statistical look at each player, followed throughout this article with explanations and discussions of each defensive stat.

Finalist Team DPS DGVT PK%  H/G B/G T/G QoC  OZ%   RCor PDiff FO%
Backes   STL  2.8 6.3 30.9% 8.3 2.6 1.8 1st 46.0%  +5.6 +0.0 48.6%
Bergeron BOS  2.7 5.7 33.3% 2.9 2.7 2.2 2nd 47.6% +18.5 +0.6 59.3%
Datsyuk  DET  2.2 3.7 21.2% 3.3 1.4 4.2 3rd 55.5% +13.2 +0.8 56.2%

Objectively determining the league's best defensive forward in a single statistic is a very difficult challenge, one that has been accepted by such pioneers as Iain Fyffe (Point Allocations), Alan Ryder (Player Contributions), Justin Kubatko (Point Shares), and Tom Awad (Goals Versus Threshold)—the latter two of which are available to us right now.

By these catch-all measurements, Backes and Bergeron rank far higher than Datsyuk, but not quite as high as LA's Dustin Brown and Anze Kopitar, who rank 1 and 2 in Defensive Point Shares (DPS) and in the reverse order in Defensive GVT (DGVT). This certainly provides ammunition to those surprised that someone like Anze Kopitar, playing on one of the league's best defensive teams, didn't make the final three. Interestingly, third place went to Gabriel Landeskog of the Avalanche (DPS) and Pascal Dupuis of the Penguins (DGVT).

A great defensive player will also kill a lot of penalties, and the fact that Bergeron and Backes killed a far higher percentage of their respective teams' penalties than Datsyuk is yet more evidence that some of the writers voted on old information this year. Not that Datsyuk should be penalized for Detroit's wealth of penalty killing options, but several forwards killed well over half their teams' penalties, like in Phoenix with league leader Boyd Gordon (54.7%) and Lauri Korpikoski (53.1%), or in Philadelphia with Maxime Talbot (54.0%).

Some pundits love to base their vote on the NHL's real-time statistics (RTSS) like hits, blocked shots, and takeaways, but there are two things to keep in mind here—these are recorded differently from one arena to another, and high totals can sometimes be as much of a reflection of the team's weakness than the strength of the player himself. After all, you can't do any of those things unless the other team has the puck in the first place.

For example, the biggest shot blockers in the league were the Islanders' Jay Pandolfo's at 6.1 blocks per 60 minutes, Edmonton's Lennert Petrell's 4.9 and Minnesota's Darroll Powe's 4.8. It certainly seems like the formula to blocking a lot of shots is to play on a team that sees a lot of them, which doesn't seem to be where you'd typically find the league's best defensive forward.

Nevertheless, Datsyuk's takeaway rate is impressive, while the shot-blocking rates of Backes and Bergeron—in the solid Talbot range (2.7) —don't rank as highly as aforementioned players like Boyd Gordon (4.3) and Pascal Dupuis (3.7). Backes' solid hitting rate puts him on par with Landeskog (8.6), Dupuis (7.1), and Talbot (7.0), but still back of Brown (10.6).

Shots Allowed

Why not just look at which player allows the fewest goals against while on the ice? Or, to remove the influence of a hot goalie, the fewest shots against? Because if you do that, all you'll find are players who get soft assignments. One of the reasons it's hard to determine the best defensive forward is because they're generally up against the top opponents, and in the toughest situations, so they consequently tend to give up a lot of shots and goals.

That leads us to an even better idea—why not just look for which player is used against the toughest opponents and in the toughest situations? While that approach will penalize players on teams with lots of good defensive options, or players whose offensive talents are even more valuable, or whose coaches aren't using them properly, it's nevertheless a useful factor to add to our equation.

Quality of Competition (QoC) is the average plus/minus of one's opponents over 60 minutes, except that it's based on attempted shots (Corsi) instead of goals. While it's not appropriate to compare it across teams in its raw form, looking at where a player falls within his own team is highly informative. Unless the team is blessed with an amazing arsenal of defensive talent, or cursed with a particularly poor coaching staff, you'd expect the league's best defensive players to be among the top-three options on their teams, which is the case for all three finalists. But again, tellingly, Datsyuk was at the tail end of that top line selection.

This metric also takes some of the wind out of the sails of those pushing for Boyd Gordon, who was Phoenix's fourth option, or even Dustin Brown, who was third. The other names we've reviewed are all still in the top three, however.

Snub      Team DPS DGVT PK%   H/G B/G T/G QoC  OZ%   RCor PDiff FO%
Kopitar   LAK  3.1 6.9 35.9%  2.5 1.8 2.6 1st 51.4% +13.8 +0.6 53.8%
Brown     LAK  3.2 6.6 28.0% 10.6 1.2 1.4 3rd 53.0%  +6.6 +1.3  ---
Dupuis    PIT  2.1 6.5 42.2%  7.1 3.1 1.3 2nd 51.7%  -1.5 -0.4  ---
Landeskog COL  2.8 5.9 23.6%  8.6 2.3 2.3 1st 54.8% +15.0 -0.3  ---
Gordon    PHX  1.7 4.9 54.7%  2.4 4.3 1.4 4th 39.8% -13.3 +0.1 56.8%
Talbot    PHI  1.3 4.9 54.0%  7.0 2.7 1.4 1st 41.7%  -2.1 +0.1 44.4%

Another strike against Datsyuk is his high offensive zone start percentage (OZ%) relative to Backes and Bergeron, and especially relative to Vancouver's Manny Malhotra, whose offensive zone start percentage is just 13.2%.

To clear up a misconception about offensive zone starts, they are calculated using only those shifts that didn't start in the neutral zone. In fact, of all Malhotra's starts, 9.5% were in the offensive zone, 61.0% in the defensive zone and 29.6% in the neutral zone (which has nothing to do with Romulans). Like most hockey statistics, offensive zone start percentage was named in a misleading way—it is more about the ratio of offensive zone starts to defensive zone starts than the actual percentage of shifts started in the offensive zone.

Furthermore, the reason Malhotra's zone starts are so low is because he is seen as an offensive liability on a team that has both elite offensive talent (the Sedins) and a coach who believes in tilting the ice in their favor. None of those factors apply to Datsyuk, or any of the other Selke finalists. Aside from Malhotra's teammates Dale Weise (20.6%), Maxim Lapierre (22.2%), and Sami Pahlsson (29.7%), other examples of such players include Winnipeg's Jim Slater (28.5%), New York's Brian Boyle (28.8%), and Chicago's Dave Bolland (32.5%).

That's not to say that offensive zone start percentages are useless in helping find great defensive players, because it's actually helpful in filtering out players who are being sheltered, and more importantly for putting their team-relative Corsi (RCor, described earlier) into context. Indeed, Daniel Sedin had the league's best Corsi at +22.5 thanks to a ridiculous 79.6% offensive zone start—contrast that with St. Louis' Alexander Steen, who had a 52.6% offensive zone start against top competition, and yet still managed a +21.7 relative Corsi (over 43 games).

Additionally, the "relative" nature of Relative Corsi could just mean that a player is on a team with a lot of weaker players making him look good by comparison—think Anaheim's Corey Perry or Edmonton's Taylor Hall.

Since it includes the ability to drive the play forward offensively, relative Corsi turns out to be a surprisingly effective predictor of who the PHWA will potentially see as a Selke candidate. This could ultimately explain Datsyuk's inclusion on the list, and certainly reflects well on Bergeron (not to mention Kopitar and Landeskog).

Extras and Intangibles

Before the advent of many of today's advanced statistics, I used defensive-situation goals (DSG) to find the league's best two-way players. Though it's based on a purely offensive statistic (goals), it includes only those goals scored in defensive situations: short-handers and empty-netters. This year, Ilya Kovalchuk was the leader with seven, followed by Jordan Staal, Zach Parise, Milan Michalek, and Mike Richards with six. Unfortunately, none of that helps us with Backes, Bergeron, and Datsyuk.

It's also important to look at the penalty differential. Certain players are highly effective at not only playing their defensive game without having to risk a lot of infractions, but also in getting their team man advantage situations, like league leaders Darren Helm and Jeff Skinner. This is certainly another factor playing in the favor of the NHL's undisputed king of drawing penalties Dustin Brown, whose extra 1.3 penalties drawn per 60 minutes means an extra quarter-goal (if the Kings had had an average power play).

One final extra intangible is the defensive advantage from winning faceoffs, especially in one's own zone. Bergeron, for example, is the league's fourth-best faceoff man, winning 59.3% of his draws, and Datsyuk no doubt prevented a lot of shots by winning 56.2% of his.

Verdict

Clearly, Pavel Datsyuk doesn't belong among the finalists this year, and earned a spot based mostly on his reputation, and for being such a big name on a great team. Anze Kopitar, the key defensive player on arguably the league's second-best defensive team, would clearly have been a superior selection.

Interesting (yet inferior) cases can also be made for players like Dustin Brown, Pascal Dupuis, Boyd Gordon, Maxime Talbot, and amazing rookie Gabriel Landeskog, who are just five of several players whose reputation no doubt earned them consideration.

Reputed  Team DPS DGVT  PK%   H/G B/G T/G QoC  OZ%   RCor PDiff FO%
Parise   NJD  2.3  5.9 37.6%  2.2 1.3 2.2 4th 54.2%  +4.2 +0.9  ---
Pavelski SJS  2.7  5.2 35.8%  1.8 3.0 2.6 1st 48.6% +11.1 +0.4 58.7%
Plekanec MTL  1.4  4.8 48.9%  2.4 2.0 1.5 1st 42.8%  -3.8 -0.8 49.1%
Staal    PIT  1.7  4.7 47.3%  4.9 1.1 1.9 1st 47.8%  +4.8 +0.0 51.0%
Marchand BOS  2.4  4.6 28.5%  3.2 0.9 1.8 1st 52.0% +17.4 +0.3 55.6%
Toews    CHI  1.7  4.0 37.0%  1.6 0.7 4.0 2nd 64.7% +14.6 +0.5 59.4%
Callahan NYR  1.6  3.7 34.2% 10.2 3.3 1.9 1st 48.8%  -2.2 +0.6  ---
Bolland  CHI  1.0  2.4 37.0%  2.8 1.8 2.5 1st 32.5%  -9.7 -0.5 48.4%
Kesler   VAN  2.0  2.9 35.3%  4.2 2.3 1.7 4th 48.0% +11.2 -0.4 53.6%
Malhotra VAN  0.5  2.6 41.1%  2.7 2.8 1.4 6th 13.0% -32.6 +0.1 58.5%

In the end, it's hard to select the league's best defensive forward, especially those whose responsibilities aren't strictly defensive, like Dave Bolland, Manny Malhotra, and Boyd Gordon, but Joe Pavelski could have potentially finished among the finalists had his team enjoyed a better defensive reputation.

There's no way to add it all up and spit out a winner, but you would expect that the league's best defensive forward would be on a great defensive team like the Blues or Kings, or perhaps the Rangers or Canucks—even the Bruins, Red Wings, and Coyotes were strong. You'd also expect them to fare well in catch-all statistics like DPS and DGVT, kill at least a third of their team's penalties, measure well in at least one real-time statistic, rank as one of their team's top-two defensive options, start in the defensive zone at least as often as the offensive zone, and still manage a strong, positive shot differential (Corsi), draw more penalties than they take, and it would be nice if they won a lot of those key faceoffs too.

There's only one player who qualifies for all those objective measurements, and that's Patrice Bergeron, our pick for this year's Selke Award. Let's hope the subjective and passionate eyes of the PHWA reach the same conclusion.

Robert Vollman is an author of Hockey Prospectus. You can contact Robert by clicking here or click here to see Robert's other articles.