Quantifying the value of your "tough" D-Man

[Steerpike]

Most every team seems to have one. That big strong defensman who doesn't handle the puck like magic and often has **** corsi% and hardly any points to speak of. Why in gods name are coaches still holding onto these guys when they could have other players who would drive possession more effectively?

I saw one study that suggested the quantitative value of tough D-Men was to limit the number of shots on net. Based on this assumption it came to the logical conclusion that these guys are worthless and an artifact due to coaches not hailing advanced stats yet.

However two things got me thinking:

1) Great offensive players have really high shooting percentages and raise their linemates' shooting percentage if they are a great playmaker. Crosby has consistently raised his linemates substantially.

2) Seeing Nate Guenin (Av's inept tough guy) battle Matt Cooke in front of Varlamov. Is the role of the tough guy really best quantified by preventing shots? I think in 5v5 situations the tough guys' moment to shine is when they are standing in front of the net and duking it out with another dickhole like Cooke. Without Guenin there to take cross-checks in the lower back some much less tough guy is there to it. This would likely lead to Varlamov getting continuously screened by Cooke and Cooke is much more free to deflect shots or take cross seam passes. Thus I think a solid tough guy, defensive D-man should manifest himself statistically by having a high on ice save percentage.

If great offensive players can get credit for their on ice shooting percentage shouldn't great defensive players get credit for having a good on ice save percentage?



Some problems to consider:

Because you usually play with a small collection of goalies, you would have to look at save percentage relative to when they aren't on the ice.

You'd probably have to correct for quality of competition to get any meaningful results as if your guy is put out there against top opposing lines that's going to lower his on-ice save percentage dramatically.

Click to read more...

 

[Freudian]

Even if they're being outshot while playing tough opposition and lots of defensive zone starts, them doing it frees up the other defenders to do other things. Just because some PMDs thrive against worse opposition they might drown against better opposition. Pairing one defensive player with a weak defensive but offensively skilled defender allows that defender to play a bigger role than they normally would. Defensive defensemen are good on the PK, where winning battles along the boards down low relieves pressure.

The idea that most defensive defensemen that don't have the best foot speed and aren't the best puck movers suddenly are worthless is an idea that's gotten traction the last year as people started equating shot attempt based stats as a perfect measure of how good a player is. It's absurd. They have a role to fill in hockey and they have a role to fill on the best teams in the league.

 

[wgknestrick]

Most every team seems to have one. That big strong defensman who doesn't handle the puck like magic and often has **** corsi% and hardly any points to speak of. Why in gods name are coaches still holding onto these guys when they could have other players who would drive possession more effectively?

I saw one study that suggested the quantitative value of tough D-Men was to limit the number of shots on net. Based on this assumption it came to the logical conclusion that these guys are worthless and an artifact due to coaches not hailing advanced stats yet.

However two things got me thinking:

1) Great offensive players have really high shooting percentages and raise their linemates' shooting percentage if they are a great playmaker. Crosby has consistently raised his linemates substantially.

2) Seeing Nate Guenin (Av's inept tough guy) battle Matt Cooke in front of Varlamov. Is the role of the tough guy really best quantified by preventing shots? I think in 5v5 situations the tough guys' moment to shine is when they are standing in front of the net and duking it out with another dickhole like Cooke. Without Guenin there to take cross-checks in the lower back some much less tough guy is there to it. This would likely lead to Varlamov getting continuously screened by Cooke and Cooke is much more free to deflect shots or take cross seam passes. Thus I think a solid tough guy, defensive D-man should manifest himself statistically by having a high on ice save percentage.

If great offensive players can get credit for their on ice shooting percentage shouldn't great defensive players get credit for having a good on ice save percentage?



Some problems to consider:

Because you usually play with a small collection of goalies, you would have to look at save percentage relative to when they aren't on the ice.

You'd probably have to correct for quality of competition to get any meaningful results as if your guy is put out there against top opposing lines that's going to lower his on-ice save percentage dramatically.

I would hypothesize that their role is not to decrease shots against, but to increase SV% due to them effectively clearing the crease, allowing the goaltender to better see the shots. Mike Weaver comes to mind here, but he doesn't fit your "tough" D man, physical mold.

I don't think it matters how "large" these guys are, but how "effective" they are. But then again, there I am wanting to make decisions based only on data and results and not "hockey logic".

I also think these guys may have some value on the PK where their ****** puck handling is not as much of a handicap because they can just ice the puck freely. It is a case-by-case issue though, and I agree that not every player that can be described by those adjectives (Tough, D man) deserves to be in the league. The other key is what game is being called by the refs? The 2006-2009 era where most holding/hooking/interference penalties were called, or the 2011-current era where the whistles are lost?

It is a great strategy to neutralize an elite F player with a scrub/goon D hanging all over them with a small chance of going shorthanded. When the refs stop calling the penalties we start seeing size trump skill. We had this "hulk" era back in the "dead puck" era. It was not a coincidence.

 

[avesfan]

The Defensive D-man is underrated. Look at the Kings lineup: they had Mitchell, Regehr and Greene, all tough, physical defensman paired with an offensive defenseman. They make shots easier, they create space for there d partner and allow them to make mistakes, and there reliable leaders, among other things... Like you pointed out, there is no stat to categorize what a defensive Dman does for one's team, but thats just a situation where old-time watch the hockey trumps advanced stats..

 

[wgknestrick]

The Defensive D-man is underrated. Look at the Kings lineup: they had Mitchell, Regehr and Greene, all tough, physical defensman paired with an offensive defenseman. They make shots easier, they create space for there d partner and allow them to make mistakes, and there reliable leaders, among other things... Like you pointed out, there is no stat to categorize what a defensive Dman does for one's team, but thats just a situation where old-time watch the hockey trumps advanced stats..

If you can't quantify it, then it probably doesn't exist anywhere but in one's mind. Mitchell and Greene have been effective and have stats to prove it (SF%, and GF%). Regehr has been bad and has stats to prove it. Remember, this isn't the "watch the game" forum. We are analytical and data driven here.

http://stats.hockeyanalysis.com/rat...00&teamid=14&type=goals&sort=PCT&sortdir=DESC

You can quantify the value of a D man pretty easily IMO, but there isn't a single item to reference. GA/20 is probably my favorite overall stat to reference along with SF% and GF% over multiple seasons. Adjustments need to be made for goaltending and team quality, ZS, etc, but that is a great place to start. I don't really like to classify them as "Defensive and Offensive" D men though. That leads you make assumptions that could cloud your analysis. Just because a D man scores points, does not make him bad at preventing goals, and vice-versa. You can't look at points and make conclusions on D play. You have to look at GA/20. You have both shot quantity and quality in that stat. It's great.

D men all have their strengths and weaknesses, but their main objective is to outscore the other team when they are on the ice (just like any player). They can complete their objective multiple ways, and most will be surprised that many of the best D men at limiting goals against (GA/20) are very "offensive" in nature. (captain obvious warning) It is much easier to prevent goals when you are playing in the other team's zone.

http://stats.hockeyanalysis.com/rat...1000&teamid=0&type=goals&sort=A20&sortdir=ASC

 

[CDN24]

The key here is that a one size fits all stats used to evaluate players is not relevant if the players have 2 different roles. Using the pens cup win as an example they had Letang and Gill on defense, by ay measure Letang is the stud of those two. They had vastly different roles as offensive big minute gu and shutdowm PK specialist. Both exceled in their role and the Pens do not win that cup without either of them. With respect to Gill the Pens have not had any playoff success since he left. Gill does not drive posession but when it came to killing penalties he excelled and was neeed. There are times in the game when you have to shut it down in your end, you better have guys who can do it.

 

[CDN24]

If you can't quantify it, then it probably doesn't exist anywhere but in one's mind. Mitchell and Greene have been effective and have stats to prove it (SF%, and GF%). Regehr has been bad and has stats to prove it. Remember, this isn't the "watch the game" forum. We are analytical and data driven here.

http://stats.hockeyanalysis.com/rat...00&teamid=14&type=goals&sort=PCT&sortdir=DESC

You can quantify the value of a D man pretty easily IMO, but there isn't a single item to reference. GA/20 is probably my favorite overall stat to reference along with SF% and GF% over multiple seasons. Adjustments need to be made for goaltending and team quality, ZS, etc, but that is a great place to start. I don't really like to classify them as "Defensive and Offensive" D men though. That leads you make assumptions that could cloud your analysis. Just because a D man scores points, does not make him bad at preventing goals, and vice-versa. You can't look at points and make conclusions on D play. You have to look at GA/20. You have both shot quantity and quality in that stat. It's great.

D men all have their strengths and weaknesses, but their main objective is to outscore the other team when they are on the ice (just like any player). They can complete their objective multiple ways, and most will be surprised that many of the best D men at limiting goals against (GA/20) are very "offensive" in nature. (captain obvious warning) It is much easier to prevent goals when you are playing in the other team's zone.

http://stats.hockeyanalysis.com/rat...1000&teamid=0&type=goals&sort=A20&sortdir=ASC

That looks a lot like plus minus stat. I disagree the objective is not to always outscore, sometimes it is just to not be scored on or limit the damage to one goal or less especially come playoff time. Defensive specialists like a Guy Carbonneau made a career shadowing guys like gretzky, his objective was not to outscore gretzky`s line, it was to limit the damage of that line to maybe a goal or less or even 2 or less so that as a team his team would win. The team objective is to outscore the opposition, the shutdown guys are there to limit the dammage so that their offensive guys can make up the difference.

 

[wgknestrick]

That looks a lot like plus minus stat. I disagree the objective is not to always outscore, sometimes it is just to not be scored on or limit the damage to one goal or less especially come playoff time. Defensive specialists like a Guy Carbonneau made a career shadowing guys like gretzky, his objective was not to outscore gretzky`s line, it was to limit the damage of that line to maybe a goal or less or even 2 or less so that as a team his team would win. The team objective is to outscore the opposition, the shutdown guys are there to limit the dammage so that their offensive guys can make up the difference.

Don't confuse the sports general objective with a coaches' answer to "How do we best achieve this objective with limited, sub-optimal resources/players?" Just because these players are in the league, doesn't mean they deserve to be or have value. If we can't quantify their value, then they have 0.

Hal Gill is by all accounts a terrible defender, until he is allowed to hold and hook his way into being effective. The fact that he won the SC does not "prove" his value, it just means that he happened to be on a SC team and didn't cost them the championship. "Limiting the damage" is accepting defeat and somewhat a means of rationalizing a failure. Can we even quantify that said defender "lessened" the damage of an elite player vs what damage we'd expect to receive from them?

We see these types of "tough" players in the league, we think "they must belong there", and never challenge their worth (whether real or imagined). The same thing happened to "tough forwards" ie fighters and they have been proven to be obsolete=valueless when challenged.

 

[CDN24]

Don't confuse the sports general objective with a coaches' answer to "How do we best achieve this objective with limited, sub-optimal resources/players?" Just because these players are in the league, doesn't mean they deserve to be or have value. If we can't quantify their value, then they have 0.

Hal Gill is by all accounts a terrible defender, until he is allowed to hold and hook his way into being effective. The fact that he won the SC does not "prove" his value, it just means that he happened to be on a SC team and didn't cost them the championship. "Limiting the damage" is accepting defeat and somewhat a means of rationalizing a failure. Can we even quantify that said defender "lessened" the damage of an elite player vs what damage we'd expect to receive from them?

We see these types of "tough" players in the league, we think "they must belong there", and never challenge their worth (whether real or imagined). The same thing happened to "tough forwards" ie fighters and they have been proven to be obsolete=valueless when challenged.

Actually if we can`t quantify their value they either have none or we do not have the tools yet to measure that value. The difference is significant. Winning the cup did not prove his value, however the next year he was a key component of the team that knocked what was essential the same pens team out in round 2.

To win a war you sometimes have to give ground. Limiting damage is not accepting defeat, they idea is to win the game. If the other team has a gretzky- Lemieux-crosby type you need someone who is skilled in neutralizing those types of players. It`s great if you have a Toews-Bergeron that can do it all- if not you need a carbonneau to shut them down and someone else to score. If the baseball analagy of advanced stats tells us anything is that you can build a team that is much greater than the sum of its parts. That`s what defensive specialists are to hockey. Maybe the measure has not been developed to accurately measure it but the coach who ignores its value will becom,e a great golfer.

 

[Aceboogie]

The Defensive D-man is underrated. Look at the Kings lineup: they had Mitchell, Regehr and Greene, all tough, physical defensman paired with an offensive defenseman. They make shots easier, they create space for there d partner and allow them to make mistakes, and there reliable leaders, among other things... Like you pointed out, there is no stat to categorize what a defensive Dman does for one's team, but thats just a situation where old-time watch the hockey trumps advanced stats..

Well, when you are paired with Doughty, its extremely easy to look good.

For instance, over the last 3 years, when Doughty and Regher play together, they have a 50.6 CF%. When Doughty is seperated from Regher, he has a 58.7% and Regher has a 43 CF% (which is horrible)

In fact, with Regher, for the last 3 seasons, over his 20 most common linemates, all but 2 (Leopold, Gaustad), have higher CF% away from him. That is pretty telling

Compare that to a smaller transitional D in Jared Spurgeon, whos most common linemates almost all do worse without him

http://stats.hockeyanalysis.com/showplayer.php?pid=1413&withagainst=true&season=2011-14&sit=5v5close

On a semi-related note, I would if there will be a gradual change in the deployment of transitional/offensive D and defensive D.

When it comes to exiting the zone, offensive D typically are much better than defensive D who have issues passing the puck and skating with it. On the flip side defensive D are much better at preventing zone entries. I wonder if coaches will start to deploy offensive D in the defensive zone more the help get the puck out, and defensive D in the offensive zone more, since the puck is in the offensive end and will either be maintained in that zone, or coming back to their own end, so deploy the defensive D to prevent this

I can see teams replacing Reghers with Spurgeons more and more in the future. With the make up of a D core 1 Greene on the bottom pairing, 2 transitional D (Spurgeron, Gardiner, Petry type guys) second pairing, and the traditional guys as top pairing D

 

[The Bob Cole]

On a semi-related note, I would if there will be a gradual change in the deployment of transitional/offensive D and defensive D.

When it comes to exiting the zone, offensive D typically are much better than defensive D who have issues passing the puck and skating with it. On the flip side defensive D are much better at preventing zone entries. I wonder if coaches will start to deploy offensive D in the defensive zone more the help get the puck out, and defensive D in the offensive zone more, since the puck is in the offensive end and will either be maintained in that zone, or coming back to their own end, so deploy the defensive D to prevent this

I can see teams replacing Reghers with Spurgeons more and more in the future. With the make up of a D core 1 Greene on the bottom pairing, 2 transitional D (Spurgeron, Gardiner, Petry type guys) second pairing, and the traditional guys as top pairing D

Don't know if I see the full value in that...just like offensive forwards, when you can skip zones and get an offensive start, you ideally want your strongest offensive players out on the ice to maximize the opportunity, as best as possible. Say you're Ottawa... Karlsson is fantastic at skating the puck through the neutral zone into the opposing teams territory and generating offensive events, but if you can get him on the ice for an offensive zone faceoff, you take it. That's why the 1OD-1DD pairing is nice to have, such as the structure LA uses. Gives you a good balance (especially if you have OD's that are relatively strong defensively too for their role) so that you can rely on someone to facilitate breakouts, yet count on them to disrupt zone entries.

But, if your team is quite awful and lacks talent throughout the whole lineup, I can see the coaches being interested in that as a 'minimizing damages' tactic. Let your best offensive players bring up the on-ice play of some of your worst defenseman, despite the fact that the D may lower the total output of your best players. Conversely, let your best transitional defensemen help out your grinders and muckers to get the puck out of your zone into the oppositions territory. This is why I think you see a lot of the middle of the pack (11-20 FF%) teams rely so heavily on PDO to make the playoffs (there was a loose evaluation of this on HEOTP that wasn't bad). You need some luck either from your goalie, or your shooters, to help get you over the hump of not having enough depth/talent throughout the lineup.

 

[Aceboogie]

Don't know if I see the full value in that...just like offensive forwards, when you can skip zones and get an offensive start, you ideally want your strongest offensive players out on the ice to maximize the opportunity, as best as possible. Say you're Ottawa... Karlsson is fantastic at skating the puck through the neutral zone into the opposing teams territory and generating offensive events, but if you can get him on the ice for an offensive zone faceoff, you take it. That's why the 1OD-1DD pairing is nice to have, such as the structure LA uses. Gives you a good balance (especially if you have OD's that are relatively strong defensively too for their role) so that you can rely on someone to facilitate breakouts, yet count on them to disrupt zone entries.

But you pair 2 PMD and the opposition will struggle containing the break out as both are strong in this regard. Pair 2 DD and the opposition struggles to break it in as both sides are strong

But, if your team is quite awful and lacks talent throughout the whole lineup, I can see the coaches being interested in that as a 'minimizing damages' tactic. Let your best offensive players bring up the on-ice play of some of your worst defenseman, despite the fact that the D may lower the total output of your best players. Conversely, let your best transitional defensemen help out your grinders and muckers to get the puck out of your zone into the oppositions territory. This is why I think you see a lot of the middle of the pack (11-20 FF%) teams rely so heavily on PDO to make the playoffs (there was a loose evaluation of this on HEOTP that wasn't bad). You need some luck either from your goalie, or your shooters, to help get you over the hump of not having enough depth/talent throughout the lineup.

I dont always tend to agree with the 1OD and 1DD theory. Its sounds really good in theory, the offensive D provides offense, the defensive D covers mistakes, but what tends to happen is the strengths become nuetralized, and the weaknesses are only partially mitigated. For instance, a team can simply choose to attack the offensive D when breaking the puck in, so they are still vulnerable as a defensive D cant cover him + cover F2 driving net. In the same sense, if the defensive D cant provide much puck moving ability, the other team can attack this player knowing he cant skate this way out of danger, and will only make a short pass, or off the boards play, and cover those exit options (leaving a forward uncovered further away)

And by puck moving D, I dont necessarily mean offensive D. I think there is an important difference in that offensive D (like Karlsson, Subban, Keith etc) can all put up points PLUS be good in transition, so when you play them in the offensive zone, you are significantly improving the chances of a goal as they can generate offensive an a elite rate. For me, PMD like Petry, Gardiner, Spurgeon etc, wont put up good offensive numbers (40+ points), but are good in transition. So putting them in the offensive zone starts wont lead to as many chances. These are the guys that should be looked at being put in more defensive zone starts. The offense for the team as a whole could actually improve, here why: Currently if they are given more than 50% of ZS, they are still only producing <35 points, so they dont really have the pure offensive instincts. You put a defensive D on the point with a bomb of a shot, and the shot generation between the two might not actually be much of a drop. However, put these guys in the defensive zone, and they are able to break the puck out much more effectively and can get the puck to the forwards in a much better position, giving them more chances to break in and score. With defensive D, they tend to struggle getting the puck out (either passing or skating), so a team is largely hemmed in to their own zone, or stuck fighting in the neutral zone for this shift


TL;DR Just spitballing ideas since its summer and Im bored.

 

[NikF]

The reason OD-DD pairings work for LA is because they use 5 man unit breakouts, the center plays really deep and almost every player on the roster is either average or better at winning board battles. It doesn't matter if the puck comes to Robyn Regehr, because 99% of the time he has an open simple pass in his zone and even if he does send it around the boards, it's a good bet that his team is winning the board battle. The team almost never tries stretch-passes and rarely if ever gets odd-man rushes (in my opinion resulting in players like Doughty leaving quite a bit of points on the board that they'd get in a different system). I've seen the likes of Matt Greene pinch and physically dominate someone in the OFFENSIVE zone keeping the puck in at the boards, because all their forwards are so defensively conscious that even he can afford to do that. When the team plays so congested they can afford to give away a bit of transition skills in a defensive-defenseman in favor of clearing the net and a physical presence. On the other hand, in a system that favors stretch-passing, or forwards leaving the zone early, or defensemen being required to skate the puck out a lot, or to play with less support, you need defensemen with mobility and quick decision making that much more. It would make sense then to favor that kind of a defenseman and give up on some of the crease clearing, boardwork and physical presence since the game you favor is probably going to result in a high pace up and down game with less stationary battles.

 

[TheDevilMadeMe]

Don't confuse the sports general objective with a coaches' answer to "How do we best achieve this objective with limited, sub-optimal resources/players?" Just because these players are in the league, doesn't mean they deserve to be or have value. If we can't quantify their value, then they have 0.

That would be true if the state of hockey analytics was such that every aspect of the sport was currently able to be quantified. Since this clearly isn't true (hockey is a fluid sport, difficult to quantify, and hockey analytics is barely out of its infancy), this is a silly statement.

 

[Czech Your Math]

I think you would have to compare like to like.

Determine which is each team's top line or scoring line(s).

Look at how different defensive pairings did against those lines (can look at Corsi, but mainly ES GF/GA or GF-GA).
The trickiest part may be trying to factor out the team's goaltending and forwards that are on the ice at the same time.

This seems the best way to get some idea of how effective shutdown D-men are.

 

[TheDevilMadeMe]

I think you would have to compare like to like.

Determine which is each team's top line or scoring line(s).

Look at how different defensive pairings did against those lines (can look at Corsi, but mainly ES GF/GA or GF-GA).
The trickiest part may be trying to factor out the team's goaltending and forwards that are on the ice at the same time.

This seems the best way to get some idea of how effective shutdown D-men are.

Right. And with GF/GA metrics, I think it's really helpful to get a sample of several seasons if at all possible.

The hardest part of all this is separating teammate effects, especially for two players who are almost always paired together.

Realistically, the only way to really do this might just be to have an "expert" watch the players play, and keep track of the number of scoring chances negated or caused by good and bad plays by the defenseman. This, of course, would be incredibly labor intensive, but I have a feeling that NHL teams already do it, maybe not for every game, but at least when scouting individual players. Once you have the good source data, you can start doing the statistical analysis.

Advanced stats are only as good as the source data going into them.

_______________________

Here is an example of what I am talking about. The author rewatched the famous 1972 Summit Series and counted scoring chances created by a player and scoring chances against created by a player's mistake. The first article was written after doing so for the first 4 games, the last after he rewatched the series.

http://blogs.edmontonjournal.com/20...m-canada-summit-series-1972-valery-kharlamov/
http://blogs.edmontonjournal.com/20...ain-of-1972-but-was-he-also-team-canadas-mvp/

This is the relevant section on defensemen:

4. Superior Canadian D-Men Were Difference
The weakness of the Soviet team was its defencemen. If a defenceman is doing his job against tough competition, he will help create one scoring chance for every scoring chance against where he makes a mistake. Usual mistakes include missing an assignment, losing a puck battle or giving away the puck.

Team Canada had two key defencemen, Brad Park and Bill White, who met that test. Park chipped in on 27 scoring chances at even strength, while making mistakes on 26 against (here are the team totals for the entire series). White contributed to 21 chances while making mistakes on 18 chances against.

The top six Team Canada defencemen were Park, +1 on scoring chances at even strength, White, +3, Guy Lapointe, -5, Pat Stapleton, -10, Serge Savard, -10, and Gary Bergman, -11, not a bad showing against such a skilled attacking team as the Soviet Union.

The Soviets didn’t have one top d-man who came close to meeting this standard.

In fact, their two top d-men, Vladimir Lutchenko and Gennady Tsygankov, helped out little on the attack at even strength, while often getting beat in their own end repeatedly.

Lutchenko contributed to 14 scoring chances at even strength, while making mistakes on 34 against. Even worse, Tsygankov contributed to 11 chances, while making mistakes on 43 against.

The top six Soviet defencemen were Valery Vasiliev, -6, Yuri Lyapkin, -12, Alexander Ragulin, -13, Alexander Gusev, -17, Viktor Kuzkin, -18, Lutchenko, -20, and Tysgnakov, with a chilly -32.

Vasiliev came on strong as the series when on (even if he made a bad giveaway on the sequence of pain for the Soviet’s leading up to Henderson’s historic goal). Indeed, Vasiliev strikes me as the first of the new breed of Soviet d-man, who was tough, even nasty, in his own end, but also skilled on the attack. This breed reached its peak with Alexei Kasatonov and Vyacheslav Festisov in the 1980s, but Vasiliev gave a taste of what was to come.

Ultimately, this is much more labor-intensive than simply using the officially recorded shot data, which is what goes into Corsi and Fenwick. But I think it's probably the only way to attempt to measure the value of a "stay at home defenseman," whose value is not in puck possession, but in his ability to reduce quality scoring chances again - by keeping goalie lines of sight clear, by keeping play to the outside (okay, this one is kind of taken into account by shot quality metrics), by clearing rebounds, by making forwards reluctant to go hard into corners for fear of being hit, etc.

 

[Czech Your Math]

Realistically, the only way to really do this might just be to have an "expert" watch the players play, and keep track of the number of scoring chances negated or caused by good and bad plays by the defenseman. This, of course, would be incredibly labor intensive, but I have a feeling that NHL teams already do it, maybe not for every game, but at least when scouting individual players. Once you have the good source data, you can start doing the statistical analysis.

Advanced stats are only as good as the source data going into them.

As I'm sure you know, I'm not a fan of subjective data.
Not saying it has no place at all, but it's best to avoid it as much as possible IMO.

I'm especially not a fan of subjective data trying to be passed off as objective or semi-objective data.

First, it's dependent on the observer. Even if the observer is unbiased (and I think most or all have some sorts of biases, whether aware of them or not), the definitions are not objective, and there's just too many ways for biases to creep into the data.

Second, there are ways in which a shutdown D-man can be effective which do not yield themselves easily to subjective viewing. For example, Scott Stevens' mere presence may prevent an opponent from seeking a higher probability scoring opportunity, but how is this going to be quantified subjectively?

So as difficult, or even impossible, as it may seem... using objective data is the only reliable route IMO. I agree that using more than one season could be helpful, but I would caution against combining the data before it's calculated (e.g., one couldn't treat '10 Ovechkin and '11 Ovechkin as "equals").