Kukla's Korner

The Puck Stops Here

Selke Trophy Leader

The Selke Trophy is given to the best defensive forward each season.  I take this to mean the forward who would earn the most win shares with his defensive play.  Generally there are two types of players who are in the race and there is no particular reason that one should be favored over the other.  There are defensive specialists who are generally third line players who have little offensive value, but are used to kill penalties and shut down the top players on the opposing team and there are two-way players who play first line roles and contribute offensively, but also play very good defence.  Usually it is two-way players who win the Selke Trophy because voters are influenced by offensive totals - even though they shouldn’t be.

I have been supporting Jeff Halpern of the Montreal Canadiens since November.  He is a third line defensive specialist who usually has little chance to win the Selke, but he should be in the middle of the race right now.  He has slowed down from a fast start and is now out with an upper body injury, so it is time to pick a new leader.

I am now picking a two-way player in Jonathan Toews of the Chicago Blackhawks.  He has a better chance of actually winning the Selke because he is a front-line two-way player and his offensive totals will impress voters.  He has earned a reputation as a top defensive forward with his strong two way play the helped his team win the Olympic gold and the Stanley Cup last year.

This reputation is deserved.  Toews plays against top competition and does a good job in shutting them down and is often able to turn their shifts into offensive opportunities for his team.  Toews is the top penalty killing forward on his Chicago team.  He is the top two-way player in the game today and is having a very good season.

I think that most likely the voters will see Ryan Kesler and Mike Richards as they are two-way players, who have scored better than Toews, but Toews has been better defensively and that is what the Selke is about.

Filed in: | The Puck Stops Here | Permalink


Chris from NOHS's avatar

Can you give us some stats on why he should be considered the best?  Otherwise, this is just a random opinion…

Posted by Chris from NOHS from Columbus, OH/Grand Rapids, MI on 01/23/11 at 09:43 PM ET

PuckStopsHere's avatar


I think you are looking for a deep statistical argument where none really exists.

I pick Toews because he is the best defensive forward when I watch him.  If you need a statistical argument he plays against one of the toughest qualities of opposition in the league - which is quite impressive because there are times you want him on the ice against weak opponents when you desperately need a goal.  While doing that he has very good puck possession numbers (for example his rate adjusted +/-).  Of players with high rate adjusted +/- ratings and quality of opposition numbers he is the only one who regularly kills penalties.

However, the real reason I pick him comes from watching hockey.  Who would you pick ahead of him?  Perhaps that is an easier question - to explain why Toews beats other candidates - of course given the vaguarities of the way Selke trophy winners are picked, I imagine more than a dozen names could come up here.

Posted by PuckStopsHere on 01/23/11 at 10:42 PM ET

J.J. from Kansas's avatar

Is it more due to the slow-down in offensive stats, the injury, or the realization that the PHWA is never going to see the Selke voting the way you think they should that brought you to this conclusion?  That’s not made very clear when you pick Toews.

Posted by J.J. from Kansas on 01/23/11 at 11:41 PM ET

PuckStopsHere's avatar


Its because I argue that his defensive play has been worth more win shares than any other forward in the NHL this year.

Posted by PuckStopsHere on 01/23/11 at 11:45 PM ET

PuckStopsHere's avatar

The fact that your Kesler argument begins by discussing the number of goals he scored is problematic.  The number of goals he scores has very little impact on his defensive play.

That said Kesler might win if the voters are impressed by his goal total.

Defensively Kesler is good, but Toews plays tougher defensive situations and does even better in them by pick possession type numbers.  It is Manny Malhotra who plays the toughest defensive situations on the Vancouver Canucks.  Toews plays the toughest defensive situations on his team.

Posted by PuckStopsHere on 01/24/11 at 12:06 AM ET

J.J. from Kansas's avatar

The win shares that you award by simply watching the games without having a system that counts everything evenly, right?

How many win shares does Toews have defensively over Kesler?

Posted by J.J. from Kansas on 01/24/11 at 12:09 AM ET

PuckStopsHere's avatar

Win shares is the idea of how to think about the situation - even if we cannot quantify it properly.

Toews beats Kesler for the reasons in my comments to Chris and to PuckHound61

Posted by PuckStopsHere on 01/24/11 at 12:12 AM ET

J.J. from Kansas's avatar

So “win shares” is nothing more than a fancy way of saying you think Toews is defensively more valuable.


Still curious about Halpern.  Is it the recent drop off or the injury that has more to do with his relative lack of “win shares” when compared to Toews?

Posted by J.J. from Kansas on 01/24/11 at 12:31 AM ET

Chris from NOHS's avatar

So basically, you just think he’s the best defensive forward is your argument. Fair enough, I have no problem with that (not saying I completely agree, but everyone has and is allowed their opinion, he’s certainly in the argument), but that makes for a pretty inconsequential blog.

Posted by Chris from NOHS from Columbus, OH/Grand Rapids, MI on 01/24/11 at 12:33 AM ET

PuckStopsHere's avatar

Halpern was dropping back to the pack from a really good start, he had likely dropped out of the lead before the injury.  I was looking at writing something like this fo a little while but the injury prompted it to be written now.

Posted by PuckStopsHere on 01/24/11 at 12:36 AM ET

PuckStopsHere's avatar

No Chris

Try reading the responce to you one more time.

Posted by PuckStopsHere on 01/24/11 at 12:37 AM ET

PaulinMiamiBeach's avatar

Win shares is the idea of how to think about the situation - even if we cannot quantify it properly.

a couple months ago we got in a discussion where you were adamant that ANYTHING you could observe on the ice could be quantified in a stat, if you chose to do so.

now you’re saying the exact opposite, just to support your own argument.

Posted by PaulinMiamiBeach on 01/24/11 at 01:27 AM ET

PuckStopsHere's avatar

I believe win shares can be quantified statistically.  We dont have the formalism entirely worked out yet.  Many people are working on the problem and I have presented several ideas that show significant progress in this idea over the years.

That said win shares are not something you observe directly in a hockey game.  They are an application of those things that you do observe directly.  The problem here is we haven’t figured out exactly how to apply those things that we do directly observe to get something that perfecly meshes with win shares.

The problem here is how to quantify just how many wins the defence of a given forward has been worth over the season so far.  We can start by quantifying how hard the role he played and how well he has done defensively in that role.  These are quantified with Toews and show that he is likely the defensive win shares leader among forwards. 

So we can turn any observation into a statistic if we wish.  The trick is turn those statistics into something meaningful that solves all potential questions we may ask.

Posted by PuckStopsHere on 01/24/11 at 01:37 AM ET

PaulinMiamiBeach's avatar

The trick is turn those statistics into something meaningful that solves all potential questions we may ask.

and therein lies your problem.  you think statistics answer all questions.  they do not.  I had hope for you when I started reading this blog post and you were talking about actually watching the games.  you dashed that hope with the above statement.

Posted by PaulinMiamiBeach on 01/24/11 at 02:11 AM ET

PuckStopsHere's avatar

And therein lies your problem.  You are attcking a strawman.  iI do not think statistics answer all questions.  That is your strawman.

Posted by PuckStopsHere on 01/24/11 at 02:13 AM ET

PaulinMiamiBeach's avatar

  iI do not think statistics answer all questions

your words:

turn those statistics into something meaningful that solves all potential questions we may ask

that is an impossibility.

Posted by PaulinMiamiBeach on 01/24/11 at 02:22 AM ET

PuckStopsHere's avatar

Right the goasl is to turn the statistics into a meaningful answer for our potential questions.  It isn’t always possible, but its the goal.

The fact that I wrote a post where a significant part of my justification for Jonathan Toews being the Selke leader is observation based and I do not have a full statistical argument and you turn around and claim I think statistics can answer all question is pretty ironic.  The post itself refutes your position.

But you did try to pull quotes out of context to make it look like I believe what I don’t believe.  Good luck in your quote mining.  It is intellectually dishonest.

Posted by PuckStopsHere on 01/24/11 at 02:27 AM ET

PaulinMiamiBeach's avatar

It is intellectually dishonest.

but it’s intellectually honest to tell me months ago that everything can be quantified, and then today say it can’t. whichever supports your argument is what applies, and this is a common theme here.

I have said repeatedly that you rely too heavily on stats.

So we can turn any observation into a statistic if we wish.

sorry, that’s not true.

Posted by PaulinMiamiBeach on 01/24/11 at 02:55 AM ET

PuckStopsHere's avatar


Once again you can turn any observation into a statistic.  Not everything can be driectly observed.

We then need a model to apply those statistics to certain problems that cannot be directly observed.  One such problem is how many win shares did the defence of a given forward produce.  We do not have a model.  We do not know with certainty what we need to put into that model.  Though we do have pretty good ideas.  Still we cannot observe directly how many shares of wins a given player’s defence creates. 

What part of this do you not understand?  I think you are fully capable of understanding it, but you don’t want to because you think you have caught me in some kind of logical error and holding onto that idea is more important to you than understanding what i am actually saying.

Want an example from another field?  We cannot measure the weight of a planet directly by putting it on a scale, but we can observe how it moves and how its moons move around it and then calculate if we have a good enough model of celestial dynamics.  If we do not have that model no data/statistics will help us observe the mass of a planet.  We can observe the periods and distances in their orbits but it won’t get us the masses without a model.  In this case we don’t have a fully formed model of how to calculate win shares from defensive values.  We can still observe and turn into statistics/data how tough the opposition a player faces, how often they score or their opponents score or shoot on goal or attempt to shoot on goal or draw penalties or giveaway or takeaway the puck or anything else.  What is missing is the model.

Posted by PuckStopsHere on 01/24/11 at 03:09 AM ET

PuckStopsHere's avatar

There are things you cannot imagine making a model to reliably measure anything regardless of what we measure statistically.

No matter how many hockey games I watch, I cannot imagine figuring out statistically the blood pressure of the players involved two years after they retire.

In fact measuring statistically how many wins (or shares of wins) a player’s defence produces is beyond my ability.  I think it may not always be beyond my ability - of course I could be wrong on that.

Posted by PuckStopsHere on 01/24/11 at 03:15 AM ET


Toews doesn’t play tough minutes, per behind the net’s corsi rel qoc. Dave Bolland does. I’d pick from a pool including David Backes, Brian Gionta, Henrik Zetterberg, Ryan Callahan, Brandon Dubinsky (it’s unfortunate that Dubi, Calli, and Anisimov are tough to separate).

In terms of GVT (our best, but far from perfect all-encompassing player-rating metric) Anze Kopitar leads in defensive GVT, with Backes 2nd, and a clump of players about half a goal behind. Not a huge margin there. Notable names appearing near the top? Backes, Nicklas Backstrom, Tomas Plekanec, Ryan Kesler, Logan Couture, Patrick Marleau, James Mirtle’s current defensive-forward-metric leader Brooks Laich, Travis Zajac, Steven Stamkos, Martin St. Louis.

Posted by Ralph on 01/24/11 at 02:50 PM ET

PuckStopsHere's avatar

As I wrote last summer defensive goals versus threshold is not a particularly reliable system.

There are multiple ways to measure quality of competition.  If we look at the line Behind the net calls qualcomp, Toews is among the league leaders (any other one works essentially the same).  The caveat on Toews is that sometimes you want him out against weak opposition so he can score.  That is what is so impressive about his numbers.  He is out in the toughest defensive situations and in offensive situations sometimes too - yet his quality of competition remains quite high.  Bolland does not play as many tough situations as Toews (or as many total minutes).  Toews also gets so easier situations when he is on the ice in offensive situations, so his average quality of competition may not be as high despite the fact he was the first choice in a defensive situation and Bolland was not.

Posted by PuckStopsHere on 01/24/11 at 04:56 PM ET


I’m not sure what you’re talking about.

Firstly, I think DGVT is a perfectly fine way to assess who we might want to consider and who we may want to leave out. It’s a starting point, nothing more.

Second, by Corsi Rel QoC, Toews is nowhere near the top. He is by QualComp, but by both metrics, Dave Bolland is higher. Toews has a 54% OZonestart; Bolland is about 35%. And Bolland only a plays a minute and a half less per game at 5on5 than Toews (14.6 to 13.1), and slighly more per game at 5on5. It seems Bolland is getting the bulk of the defensive responsibility.

Look, I don’t doubt Toews is much better than Bolland defensively, nor that Toews is Selke-good defensively. But is he carrying the load? No. He may be able to get it done defensively, but he isn’t getting the ice time to do it. He’s being deployed in offensively-advantageous situations because that’s the best use of Toews’ abilities on the Blackhawks. While it may seem unfair for me to assess him like that since he’s not controlling his ice time, that’s hockey. I don’t think Selke winners should only be secondary defensive players.

I forgot Eberle before. While I think he’s been with Horcoff a lot, as a rookie what he’s done is mighty impressive, beating the toughs. Should be in the Calder conversation and might be a Selke winner pretty soon.

Posted by Ralph on 01/25/11 at 01:11 AM ET


Just a bit more on DGVT…that’s not to say I think it gives us our complete, unedited preliminary consideration list. It just gives us names to take a closer look at, and there are names missed, obviously. Just that deserving guys flying a bit under the radar may make a showing there (guys like Jay McClement, perhaps) and hence we take notice even without MSM coverage of the guy.

Posted by Ralph on 01/25/11 at 01:13 AM ET

Add a Comment

Please limit embedded image or media size to 575 pixels wide.

Add your own avatar by joining Kukla's Korner, or logging in and uploading one in your member control panel.

Captchas bug you? Join KK or log in and you won't have to bother.


Notify me of follow-up comments?


Most Recent Blog Posts

About The Puck Stops Here

imageThe Puck Stops Here was founded during the 2004/05 lockout as a place to rant about hockey. The original site contains over 1000 posts, some of which were also published on FoxSports.com.

Who am I? A diehard hockey fan.

Why am I blogging? I want to.

Why are you reading it? ???

Email: y2kfhl@hotmail.com