About last night… (and about the Leafs holding leads)

Advertisement - Continue Reading Below

“Score effects” is the term used by hockey analysts to explain why teams that are behind by a goal tend to out-shoot the opposition. You may notice in the third period, often a team will rally but come up just shy. I’ll point to examples last night, such as Carolina, leading 4-1 into the third period against New Jersey, were out-shot 9-3 and held on for a 4-3 win. Winnipeg, up 3-1, were out-shot 10-7 and won 3-2. Phoenix was up 2-0, was out-shot 14-11 and won 3-1.

Not to say this happens every game, but comebacks happen. We don’t know why, but in almost any sport, the trailing team generates some artificial momentum and does better.

That doesn’t explain what’s up with the Toronto Maple Leafs when they attempt to hold a lead. Nevermind getting out-shot 19-0 by the Penguins in the third period and overtime after being up 5-3 going into the final 20, the last time the Leafs held a lead going into the third, against the Islanders, they were out-shot 17-10. (I incorrectly wrote that the Leafs had been out-shot 36-4 in third periods their last two times going in with leads in a comment on the postgame thread. It’s really 36-10. Misread something).

I don’t want to re-hash too much what happened against Pittsburgh, but I want to point you to this chart here. We know that the Maple Leafs are a sub-50% puck-possession team. This is measured by Corsi, a metric that tracks every attempted shot at either end. 50% is even, and so on.

Advertisement - Continue Reading Below

This is breaking down last season’s statistics. I looked at the NHL median for each 5-on-5 game state (tied, up by 1 goal, up by 2 goals, etc:) and noticed something fishy:

  NHL Median Leafs % Leafs Rank Diff
Up 2 43.0% 38.1% 26th -4.9%
Up 1 45.8% 40.5% 27th -5.3%
Tied 49.7% 43.8% 29th -5.9%
Down 1 53.5% 50.6% 25th -2.9%
Down 2 56.7% 54.9% 26th -1.8%

(Numbers via Hockey Analysis — Note how the NHL median Corsi % increases as you subtract a goal from the net total. The average team does better Down 1 than Tied, for instance)

Advertisement - Continue Reading Below

In the Up 2, Up 1 and Tied game states, the Leafs are much further away from the median, between 4.9% and 5.9%, compared to 1.8% and 2.9% when they’re down by 1. This could be coincidental, but a similar discrepancy comes this season:

  NHL Median Leafs % Leafs Rank Diff
Up 2 44.9% 37.2% 27th -7.7%
Up 1 47.1% 40.3% 28th -6.8%
Tied 49.6% 41.9% 28th -7.7%
Down 1 53.1% 51.5% 21st -1.6%
Down 2 53.0% 43.0% 30th -10.0%

(Keep in mind, the Leafs haven’t played a lot of minutes down by two goals, and when they have, it’s been against Chicago, Vancouver, Boston… some pretty good puck-possession teams. And also Columbus)

Advertisement - Continue Reading Below

I want to concentrate on that “Down 1”. The Leafs are much, much closer to the NHL median when they’re down by 1 than when they’re Tied or Up by a goal. Tyler Dellow at Sportsnet also noticed this trend when looking at individual Corsi rates for Dion Phaneuf, in a post well worth your time:

His Corsi% when tied and leading has, however, absolutely collapsed relative to this group of players. That’s going to lead to fewer shots when he’s on the ice and fewer points—and we’ve seen from his numbers that that’s the case.

First, perhaps Phaneuf has experienced some sort of a decline that affects him only when the Leafs are leading or tied. Second, it’s possible that the Leafs have changed the way they play when they’re leading or tied, and that has disproportionately affected Phaneuf.

Why are the Leafs 5-7 percentage points below league median when leading or tied, but are much, much closer to the median when they’re behind? There has to be some strategy at play here, something that Randy Carlyle is doing that’s hindering the Leafs ability to generate and prevent shots. I’m no scout, but it doesn’t take a genius to note that there are some real problems with the breakout, and that Leaf goals are created almost exclusively off of counter-attacks. Against Pittsburgh, the Leafs scored their goals in each instance within seconds of recovering the puck off a Penguin player’s stick.

That only really works when you’re getting great goaltending. Leafs goaltending has really yet to falter this season, but to paraphrase Jonathan Willis from Bleacher Report, both James Reimer and Jonathan Bernier are getting numbers similar to Dominik Hasek in his prime, and barely flirting with 500 in the last 17 games.

If the goaltending holds up, great, but there exists a solid discrepancy between how the Leafs play when tied and when down by 1. If they had the same 2% difference compared to the median when tied, rather than 7, that’s about 12 fewer shot attempts per game that the goaltenders have to face, and 6-7 fewer actual shots. I don’t know what it is, but somebody’s going to have to figure it out soon, because leads are suddenly coming back to cost the Leafs a little bit more.

And now to look at the Pittsburgh game again

Now… about the Pittsburgh game. Here’s the Fenwick Chart, noting unblocked shots over the course of the game:

It’s actually pretty close. The Penguins held just a two shot advantage going into the third period and OT, where they went up 29-3. The bad call on Jerred Smithson led to a powerplay goal against and Evgeni Malkin got away with shoving Jonathan Bernier into the net, but I don’t think that the Leafs really deserved to hang on in this one. As ugly as the Penguins played in the first and early second, it was reversible. There were some bad giveaways, but ultimately possession trumps all that, and the Penguins came in wave after wave against Jonathan Bernier, who played great.

Some notes, because I didn’t have any in the post-game thread. Numbers via Extra Skater’s game sheet:

  • Nazem Kadri started six shifts in the offensive zone and zero in the defensive zone, yet the Leafs were out-shot 4-13 with him on the ice and out-attempted 13-22.
  • Kadri played a lot against Evgeni Malkin, but Malkin was surprisingly one of the worst Penguins by way of puck possession on Wednesday. The Penguins out-attempted Toronto 23-19 with Malkin on the ice. He was mostly matched up against Morgan Rielly and Paul Ranger, and those two were the only regular Leafs over 40%.
  • I found that pretty interesting, but Rielly-Ranger weren’t really trusted by Carlyle to start shifts. The Leafs had to take 24 shifts in the defensive zone, and while Dion Phaneuf started 14 and Carl Gunnarsson 13, Rielly had 3 and Ranger 4.
  • Jerred Smithson had 5, somewhat surprising given his role usually, but he wasn’t having a great night in the circle, at just 4-for-10. The Leafs didn’t record a single shot on goal in the 9.2 minutes Smithson played.

Ultimately, this is a game that the Leafs are going to want to burn, but the problems regarding holding a lead persist. This is somewhat in response to a commenter in the postgame pointing out that there’s no real empirical proof that this is Carlyle’s system at work, and instead a problem of roster composition.

I’d agree, I don’t think the numbers I have are a strong-enough argument, but it’s worth asking why the bulk of Leafs possession troubles seem to occur when the score is tied or when the Leafs are leading. If the answer turns out to be “luck” and by March, these numbers sync up, then at least we didn’t say we didn’t ask the question.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

  • STAN

    Good little analysis Cam. Looking at last night in a vacuum, I don’t usually like to do this but the refs really dropped the ball and it really threw of the Leafs. The Smithson and Holland calls were so so soft, and the Malkin non-call goal was the icing on the cake. This doesn’t excuse their Terrible third period, but does help to explain a little bit as they were pretty good the first two frames. I’m sure the team feels like they just can’t catch a break right now, but they need to find a way to battle through.

  • Back in Black

    Refereeing mistakes are frustrating, but they do happen to every team eventually. Call it bad luck – but the Leafs had a lot of good luck in this game too, as the Penguins essentially gifted them three goals.

    Is it Carlyle? IIRC there is evidence that he was not a very effective coach for most of his time in Anaheim, considering the personnel.

    Is it player composition? Well, Carlyle bears responsibility for problems both in the lineup (one of his best defencemen was inexplicably in the press box last night, while three barely-AHL-quality players were on the fourth line) and in the roster (he drove two of his best possession forwards off the team in the off-season).

    • Back in Black

      Luck is one thing, but this is the first game all year that I can.say the Referees had a big hand in. Teams can work hard and create their own luck, as Bozak did in Letang for Kessel goal.

      No denying Carlyle is a good coach, he’ll get it figured out. This team has the talent, they are just in a little slump. Gardiner had been.one of their worst dmen lately and deserved to sit. Hopefully it woke him up. I do wonder though if the Leafs conditioning could use some help. Many players seem to gas out quickly and later in games. Tough to address this now though.

      Sure the Leafs lost two good players, but the replacement of Clarkson for Maca is an upgrade, as Clarkson brings a Mich needed cycling precense to that lineup. I’d also take Bolland in a 3rd line role over Grabbo. This team really mosses Mr. Bolland. Bozak has also been good this year, but most people are blind to this and like to use him as a Scapegoat.

  • STAN

    I think that Tyler Dellow article is “plagerised” or at least loosely based and unreferenced version of David Johnson’s work last year on Phaneuf – LINK

    Johnson works is much more thorough and the analysis presented allows the reader to make better conclusions.

    Second even with Tyler’s Calgary only data Phaneuf looks like an “average” NHLer but which doesn’t account for his “better” team mate in Calgary nor the QOC that he faces. This is poor and misleading analysis on his part.

    Edited to include Link –Ed

    • I don’t know. I read that thing from David and have quoted it here before, but Tyler was the first to look at the data from the Leafs as a whole and conclude it may be an overall strategy thing.

      Obviously I’d love to see David weigh in here.

      • STAN

        I think coaching and systems do affect corsi which is why it starts to lose some of its predictiveness over the long haul. I think there needs to be more effort into determining how much a players corsi can be affected by systems but that isn’t an easy thing to do.

        As for Phaneuf, his ‘trailing’ stats indicate that he might be a good or maybe a very good defenseman (I’d hold of calling him elite), particularly his trailing stats of a few years ago. The problem is, he is completely mis-cast with the Leafs. His usefulness isn’t in a low-risk conservative system as a shut-down defenseman. His strengths are in the offensive zone, running the offensive play, keeping the puck in the offensive zone, occasionally approaching the net to catch the defense off guard, etc.

        My question about Phaneuf is, is it better paying him big bucks to play a style he isn’t suited for like he is doing now or do you pay him big bucks and switch him back to a more offensive game that he is more suited for. If it is the former, that seems stupid to me. If it is the later, I have to wonder if Gardiner and Rielly might be better and cheaper options for that role and then use a portion of Phaneuf’s $$ to get a better shut-down defenseman. So, that leaves me wondering if signing Phaneuf to an 8 year big bucks contract makes any sense.

        • Jeremy Ian

          This is a very good thread.

          Cam’s right — if you want to win games, more wins are going to come from holding the leads; Cam’s numbers make me worry.

          David’s point about Phaneuf’s placement is insightful, and does raise the issue of a coach’s system. I guess I’d have more faith in the decision to exchange Phaneuf for a cheaper shut-down defenseman if I had more confidence in the coach’s system.

          Which raises the issue that I think David is getting at, which is that it’s hard to draw a hard line between the system and player composition. (Back in Black’s point). Because the composition, especially in this offseason, was calibrated to the coach’s system.

          So, ultimately, to me this string of games is pulling back the veil on Carlyle. I am willing to withhold a final judgement. But the prosecution is winning this case so far.

          We can blame bad luck (ref’s — yes, they were awful last night; injuries — but what about Columbus’ limping roster?), but there’s a pattern developing…

        • I’ve been thinking the same thing as well.

          It’s ironic because with the Leafs adding Gardiner, Franson, Rielly in place of Komisarek and Beauchemin, they were essentially inviting a faster game, but whatever it is the team is doing is taking away from the strengths of the individuals in place.

          I still think John-Michael Liles has the talent to be an effective defenceman.

          • Jeremy Ian

            I don’t disagree on Liles, but he just doesn’t fit in on a team with that has Phaneuf, Gardiner, Franson and Rielly as well.

            The fact of the matter is, if you just want a defenseman to play purely passive defensive hockey like they are asking Phaneuf to do much of the time you can get them a lot cheaper than what Phaneuf will sign for. That’s effectively what Mark Fraser is and I am not sure there is a lot of evidence that Phaneuf is all that much better at purely passive protect-the-lead defensive hockey than Fraser is.

  • By the way, I looked at the same thing for Ron Wilson, in the Leafs’ last year. Obviously the “Up 2” is corrupted by the fact the Leafs didn’t often have a lead like that, but Wilson’s problems appeared to come when the game was tied. Otherwise, the Leafs were closer to league average in terms of possession:

    Chart here

  • STAN

    @David Johnson

    Great points about Phaneuf. But my eyes also tell me that he is far slower than he was during his Flames seasons and perhaps the first season in TO. That happens when your skating style relies on one leg for power (he constantly pushes off one foot to pick up speed).

    His lack of jump is at the root of most of his problems, I think; forwards getting around him and shielding the puck, beating him to pucks that have been softly flicked into a corner, and trying to chase down pretty much any forwards on a two on one.

    He’s never had soft hands, choosing to pound the puck around the boards, usually to a pinching defenceman or forward who has done his homework and KNOWS what’s coming.

    BUT, that comes straight back to coaching. Either Carlyle leaves him alone because he’s ‘Captain’ and considered elite (which he is not) or he keeps trying to improve him and is being ignored. Either way, the Leafs have a problem.

    I think they’d be far better off shopping him NOW for something and let him ride out his career somewhere else.

    A telling Carlyle quote after last night’s debacle in Pittsburgh went like this, “We didn’t skate and didn’t forecheck or establish much…there’s no explanation for us not getting any shots in the third period when we have an old enough group and a veteran corps that should be able to grab a hold of it.”

    Seems to me he’s admitting he has no plan and no influence with the so-called ‘core’ group, which would start with Captain Phaneuf and include the likes of Kesssel, Lupul, Clarkson and Bozak (based on pay-cheques, which are supposed to reflect value and leadership.)

    It’s MORE frustrating than the last two seasons of Wilson’s rein.

  • STAN

    Also, when it comes to the pure mobility of Leafs defensemen I’d rate them…

    1. Rielly
    2. Gardiner
    3. Liles
    4. Franson
    5, Phaneuf
    6. Gunnarsson
    7. Ranger
    8. Fraser