Dion Phaneuf and Jake Gardiner, comparing matchups

I might need to preface everything I say about Jake Gardiner thusly: I think he’s a talented, exciting defenceman who will be a good Top Four option in the NHL one day. The other thing I need to say is that it just won’t be now.

Young defencemen, particularly those who score a few pretty goals (such as the one after the jump), have a way of making us dream. The harsh reality is that defencemen develop a little slower than forwards and their actual impact on an NHL team won’t be felt for some time.

Again, here are Gardiner’s advanced numbers for this last season via Behind the Net:



  TOI/GP Corsi ON Rel QoC Ozone%
Jake Gardiner
(Team rank out of 7)
17.11
(3rd)
0.28
(2nd)
-.328
(6th)
56.0%
(7th)

(Team rank doesn’t include Jay Rosehill)

“Corsi ON” indicates that when Gardiner was on the ice, the Leafs had 0.28 more shot attempts (shots on goal, misses and blocks) per 60 minutes directed at the other team’s net than against. That only includes 5-on-5. But as the other two numbers show, this is not as impressive as it looks. For starters, Gardiner started 56% of his shifts in the offensive zone and played the second weakest competition the opposition had to offer, so we’d expect positive possession from Gardiner.

He does do this:

But it’s at the expense of the Leafs putting a No. 5 defenceman in the No. 3 slot. It’s good that the team has found a lot of easy minutes for Gardiner to play, but that puts a tremendous burden on the top pairing for the Leafs.

An article written by Eric. T at our sister website NHLNumbers.com yesterday on the real impact of quality of competition numbers, concluding the following:

These competition metrics provide valuable insight into what a coach thinks of a player and how he tries to use them, but in practice they do not show differences large enough to have significant impact on the player’s results.

What, for instance, is the difference between Gardiner’s -.3 Rel QoC and Dion Phaneuf’s +1.4 Rel QoC? Eric suggests that “such small corrections are scarcely worth the effort” but I disagree with this in principle.

Without taking quality of competition into account and correcting Corsi for only zone starts, that would lead one to the conclusion that the Leafs were better with John-Michael Liles on the ice than with Dion Phaneuf. That’s despite Phaneuf being on the ice for many of the tough match ups.

Consider Jason Spezza, who I consider to be the top offensive talent in the Northeast Division. The Leafs faced Spezza’s Ottawa Senators six times this season. Splitting up how much time Spezza played against Phaneuf, Gardiner or Liles is critical, as well as their third- and fourth-line centre men, Zack Smith and Zenon Konopka:




Spezza Smith Konopka
Phaneuf 61.0 16.0 5.6
Gardiner 15.9 29.6 11.2
Liles 20.4 16.3 6.6

You can see here that Phaneuf played an hour of even strength time against Spezza, compared to Gardiner’s 16 minutes and Liles’ 20. This is tremendously important when deducing a player’s impact. Gardiner played the majority of his Ottawa minutes against Zack Smith, Phaneuf played against Spezza.

The Senators have the puck 50.2% of the time when Smith is on the ice without Gardiner. The Senators have the puck 53.4% of the time with Spezza on the ice when he isn’t matching up against Phaneuf. I think the distinction is important, perhaps not so much in telling whether a defenceman is a good two-way talent, but in determining his overall level of importance to the team.

Eric does go on to say that these match ups do determine Corsi rates over fairly small samples, like shot quality, they’re determined over the span of a game, or six, but if the coach is successful at getting the best match ups for his team, I think you’ll see a much larger effect. A lot of the top offensive talent isn’t necessarily the best possession talent:



2012 Spezza Smith
Goals For 3.6 2.49
Corsi 9.43 1.11
     
2011 Spezza Smith
Goals For 3.39 1.38
Corsi 2.58 3.15
     
2010 Spezza Smith
Goals For 3.1 1.53
Corsi -1.4 -2.04

In every year of the Behind the Net era, Spezza is dominant when it comes to “goals for per 60 minutes” but not Corsi. He’s a little more average there and was even beaten by Smith in 2011. Over a run of years like this, I’m quite certain that the shots taken by the Ottawa Senators with Spezza on the ice have a much better chance of going in than when Smith is.

David Johnson at Hockey Analysis frequently uses goals for and against in his analysis. Here at the Leafs Nation and many other blogs that frequently use advanced analytics, we tend to use Corsi rates a little more often because there are more Corsi events than goals. I think a mix of the two is required, goals for on offence, Corsi on defence, to determine how strong a player truly is. 

For Gardiner, as much as I like him, his ability to drive play is lacking, and he doesn’t play against top offensive talent. The first half of the season will be very important for him in establishing exactly where he fits in the lineup. I don’t think the Leafs will continue to be able to get him 17 minutes a night against the third and fourth lines of the world, which is what makes him successful.

He will improve this season, I’m sure, but by how much? The Leafs have a No. 1 defenceman, but after that, they have a No. 3 playing as a No. 2, a No. 5 playing as a No. 3 and so on. They still need to acquire another top talent to push everybody down the depth chart and into the minutes they can appropriately handle as NHL defencemen.

  • Danny Gray

    I read something recently (and feel some shame that I can’t remember where it was from so I can source it) that QoT seemed to have a greater impact than QoC.

    If that’s the case, does that not make Gardiner look quite a bit better than the numbers above? I may be wrong on this but I think his QoT was abysmal last season.

    • I think QoT has more of an impact than QoC for sure. It’s just that QoT is so circular. By generating a Corsi event for, you’re impacting the Corsi scores of all five opponents, and hence impacting your own QoC. But you only see the same opponent, at most, for ~5% of your ice time. A consistent line might play 1/3 of the time together, so during that time each Corsi event gets counted double for QoT calculating purposes. You don’t have the benefit of a huge sample of teammates to, more or less, drown out that circularity, like QoC has.

  • Not Norm Ullman

    The biggest flaw in this analysis… Phaneuf was partnered with the Leafs’ best defensive d-man most of the year. Gardiner was usually partnered with their worst defensive d-man.
    The overrating of Phaneuf must stop. He is NOT a legitimate #1 d-man. On any team.

  • I was surprised during Burke presser he said the areas that the leafs needed to improve are 1C, Goaltending and size. No mention of the need to improve defence. And no clue what they are missing because GA (and even shots against or corsi) show that the problem is not just goaltending.

    I tend to think – management and Burke feels the problem is goaltending and not both goaltending and defence.

  • Great article. The Leafs are in an interesting situation due to what you mentioned above. But they also have guys who play tough minutes playing on their offsides. Phanuef is better playing on the let (his natural side). But they didn’t have another ‘tough mins’ d-man to play the right side, so he shifted over while Gunnar stayed on the left.

    In todays NHL, not many players can effectively play on their offside. The games to fast. Parise said they specifically targeted this weakness in the Flyers defense – as they also had to many guys playing on their offsides.

    Anywhoo, interesting stuff.

  • Not Norm Ullman

    I don’t get it, Cam. I’m happy to agree that QoC matters. Fine. Also, Gardiner got sheltered. As he should have – being a rookie right out of the NCAA, and one of the lightest defencemen in the league.

    But. He also played with a terrible set of team-mates. What is the point of not showing his QofT? I’ll repost them here from the column the other day, along with a bit of text:

    “Gardiner had really weak team-mates, with a Relative QofTeam-mate of -0.516. This was the weakest set of team-mates amongst Leaf defenders, and the 15th weakest set of team-mates of all defenders (over 70 games) in the entire League.

    Gardiner -.0.516
    Dion -0.043
    Liles +0.212
    Schenn +0.257
    Franson +0.389
    Komisarek +0.682
    Gunnar +0.730

    And again, I’m happy to consider Corsi. I think it’s important. But why not also consider Goals? I mean, immediately below the Gardiner discussion, you bring them in to discuss Spezza…. but in an article nominally about Gardiner, you exclude them?

    Which is odd, because he has easily he highest Goals For per 60 minutes ice-time of all Leaf defencemen, and amongst the highest in the league. e.g. Jake was 2.95 versus Dion at 2.21 and Lilies at 2.03.

    And again, in terms of Goals For in 5 versus 5 Close situations, Gardiner played great, generating almost TWICE as many goals as Dion, Gunnar or Liles.

    Gardiner 1.165 -> the 2nd best in NHL for defenders >500 minutes Schenn 1.03
    Gunnar 0.694
    Dion 0.693
    Liles 0.559

    And to compare with the rest of the league:

    Keith 1.107
    Karlsson 1.105
    Lidstrom 0.966
    Chara 0.900
    Weber 0.858
    Doughty 0.661

    And again, as per the other day, even on DEFENCE, in these 5v5 Close, zone adjusted situations, Gardiner was better than Gunnar, and not that far from Dion (the best defender.) In a league-wide comparison, he also did quite well – better than many big names.

    So what we’re seeing here is that MOST of the seeming failure of Gardiner as a defenceman came in situations where the Leafs were trailing, and he was being called on to try and salvage the situation – i.e. play run and gun. Here are the figures, goals against per 20, 5v5, Close, zone adjusted:

    Dion 0.773
    Liles 0.804
    Gardiner 0.858
    Gunnar 0.875
    Schenn 1.163 (3rd worst in NHL defence playing >500 such minutes)

    And we see Keith 1.107 Karlsson .903 Keith .992 Chara .852 Lidstrom .773 Doughty .635 Weber .572

    So, again, I reach the same conclusions. Absolutely, Gardiner has to improve on defence. Absolutely, he got a boost from positive zone starts, and absolutely, he faced weaker competition.

    But. He did so with weak team-mates. And yet, when the game was close, he scored like crazy, while playing not bad defence.

    And mostly, what we see as his “bad defence” turns out to have been on NON 5 v 5 close situations – i.e. when games entered their late stages, and the Leafs were well behind, and it fell to him to try and run and gun – that more goals against got scored.

  • Danny Gray

    “I think a mix of the two is required, goals for on offence, Corsi on defence, to determine how strong a player truly is.”

    I am willing to accept this. I do think skaters can influence shot quality against, but it is difficult to quantify and probably pretty minimal for the majority of players.

    If we sort Leaf defensemen by 5v5 ZS adjusted OppGF20 (which is the average of all opposing players GF20 weighted by TOI against) we see the following:

    Phaneuf: 0.823
    Gunnarsson: 0.814
    Komisarek: 0.789
    Liles: 0.782
    Gardiner: 0.780
    Schenn: 0.767
    Franson: 0.763

    (See http://stats.hockeyanalysis.com/ratings.php?disp=1&db=201112&sit=f10&pos=defense&minutes=500&teamid=28&type=goal&sort=OPPF20&sortdir=DESC)

    Clearly Phaneuf and Gunnarsson were use against the toughest opponents offensively and Franson and Schenn were somewhat sheltered.

    Phaneuf and Gunnarsson’s OppGF20 are on par with Regehr (0.820) and Sekera(0.818) in Buffalo but below that of Gorges (0.839) and Subban (0.836) in Montreal. Surprisingly Chara’s OppGF20 is only 0.808 but that is in part due to the fact that he didn’t have to play against the Bruins who were clearly the best offensive team in the NE division. Strangely Kuba(0.803) had the toughest OppGF20 in Ottawa. It doesn’t seem Ottawa employed a defense matchup system I guess in large part because they didn’t have a reliable shut down tandem.

  • There are some definite problems with Phaneuf playing with Gunnarsson and Gardiner playing with Schenn, but it doesn’t affect how much time they spend against forwards.

    The problem with QoT is that Gardiner’s teammates also don’t face the same quality of minutes. Looking at Gardiner’s WOWYs would lead to assume that he was much better without Schenn: http://stats.hockeyanalysis.com/showplayer.php?pid=1513&withagainst=true&season=2011-12&sit=5v5&type=corsi

    It’s just that Schenn, without Gardiner, had much tougher minutes. If we split the difference (Schenn had a -.14 something QoC and a 50% OZ rate) since they played a similar amount of minutes both with and without, that gives Schenn a .2 QoC and about a 44% OZ rate without Gardiner, where he would be expected to be awful.

    Gardiner’s teammates got dragged down by tough minutes when they were without him. It makes it very tricky. It didn’t affect how much time Gardiner saw with the Leafs forwards like Grabovski or Kessel.

    • “Gardiner’s teammates got dragged down by tough minutes when they were without him.”

      I am not sure I can accept this. Gardiner’s most frequent defense partners were Schenn and Franson and by almost any metric those two were the most sheltered Leaf defensemen. Schenn and Franson were the only two Leaf defensemen that had a lower ZS adjusted 5v5 OppGF20 than Gardiner (i.e. when not playing with Gardiner they played against even weaker opposition). Gardiner also had the lowest OppFA20 indicating no Leaf defenseman played against tougher defensive opponents while Franson and Schenn were 5th and 6th (ahead of only Komisarek).

        • Not Norm Ullman

          Phaneuf-Gunnarsson definitely top pair. Gardiner I think deserves to be on second pair but I don’t know who goes with him. A nice 2-way defenseman would be good but the Leafs don’t have a second Gunnarsson. There are no good options really. If Carlyle can refocus Komisarek maybe Komisarek is the guy. I am crossing my fingers that Komisarek does better in a more defensive oriented system. If he wanted to make it a more pure offensive paring Franson could be the guy. The third pairing is Liles and whoever doesn’t play with Gardiner. Liles and Franson actually played fairly well together last year so they might be a good fit. Holzer will likely be the 7th defenseman but depending on how well Franson and Komisarek play (and injuries) might get a fair bit of time in the line up.

          If I had my guess, the pairings would be Phaneuf-Gunnarsson, Gardiner-Komisarek, Liles-Franson with Holzer as 7th defenseman. Blacker could get some time with the Leafs too as an injury fill in.

  • Not Norm Ullman

    @Cam. Actually, when you look at the WOWY page you linked to, it pretty clearly shows that not only did Gardiner’s performance fall when paired with Schenn, but it soared when paired with Dion.

    Even more important, Gardiner dramatically raised the performance of ALL of the higher-end Toronto players he was paired with, from Dion (who went from 49% chances for when apart from Gardiner to 61% chances for with him)… through the Kessel line (up by 3%)… and even moreso, with the Grabbo line (which went from 50% chances for to 56%.)

    There is far more positive in these stats than negative. Gardiner may have a down year, but I’m not convinced it’s because he’s really bad at defence.

    Also, when it comes to offensive defencemen, their development curve is a bit different I believe. That i, their offense tends to show up relatively quickly, while learning defence – as for most defencemen – takes longer.