33
Photo Credit: Anne-Marie Sorvin - USA TODAY Sports

Beware of what a defenceman’s points are(and aren’t) telling you


Points totals aren’t always entirely reflective of a defenceman’s offensive contributions.

As counterintuitive as this sounds, it begins to make more sense when you examine the fundamental reason as to why we care about them in the first place.

People value points because we’re led to believe that the more a player collects, the better he is at helping his team score goals. While that’s proven to be true for forwards, there doesn’t actually appear to be much of a correlation for defencemen.

A quasi-study conducted by Travis Yost found that there isn’t a meaningful connection between the rate at which a defenceman generates shots and points and a team’s scoring rate with that defender on the ice.

In other words, your team won’t necessarily score more just because a point producing defenceman is on the ice. It sounds silly, but Yost’s reasoning for why this paradox exists doesn’t sound unreasonable.

Yost found that a defenceman’s goal and assist totals are tied closely to the rate at which they shoot the puck. This contrasts forwards, who only see a relationship between shots and goals. In simpler terms, it means that the more a defenceman shoots, the more likely they are to get points.

Based on this, you might be duped into thinking that it’s in the teams best interests for defencemen to shoot the puck. But as we mentioned at the start, there’s no correlation between the rate at which a defender registers shots and the scoring rate of his team while he’s deployed.

A reason for this missing link is because there’s an opportunity cost every time a player attempts a shot. Every time a player shoots, he’s forgoing the chance to make a pass that could eventually lead to a more dangerous shot.

Opportunity cost is a lot more relevant for defencemen. Shots from the point are largely inefficient as they have just a 3% chance of going in. In most cases, you’d be better served as a blueliner to make a pass to keep the play alive rather than fire a long-range shot.

The problem as you’ll recall is that the existing points model rewards the opposite – defencemen pick up points in relation to their individual shot volume.

As Yost mentioned in his article, this often leads to situations where defencemen accumulate vacuous assists.

What should we be using to evaluate a defenceman’s offensive contributions?

If points alone aren’t enough to measure a defencemen’s offensive contributions, what else should we rely on?

Transitional play sticks out as an overlooked component of offence. A defenceman who’s able to move the puck out of the defensive zone and/or into the attacking area with possession creates a territorial scoring advantage for the rest of the team, even if that player doesn’t receive credit in the form of points.

Unlike a defenceman’s individual points rate, effectively moving the puck up the ice is statistically proven to improve the team’s overall scoring proficiency. Zone exits with possession improve the rate for possession zone entries – the latter of which are nearly twice as effective at producing shots and goals than dump-ins.

In terms of offensive zone play, I tend to lean on Ryan Stimson’s data. Stimson’s passing project tracks passing sequences in an attempt to measure pre-shot movement – a critical, yet underrated part of scoring goals.

At a team-level, his passing model is one of the most powerful public tools – more predictive for future scoring than shots, goals and other expected goal models.

For defencemen, primary shot contributions(primary shot assists + shots), as well as Stimson’s expected primary points per hour metric, are better at predicting future primary points than a player’s past primary points.

Explained more simply, primary shot contributions and expected primary points are better tools at predicting a defenceman’s future point totals than past points themselves. This means that we can rely on those two metrics to identify players that have counting stats that aren’t commensurate with their true offensive contributions.

An example of this that Canucks’ fans can relate to is Ben Hutton. The former fifth-round pick led Vancouver’s blueline in scoring with 25 points in his rookie season, even though Stimson’s expected primary points metric pegged him at a bottom 20th percentile rate for the 2015-16 season.

Since then, Hutton’s point production has fallen in consecutive seasons to align closer to the scoring pace that the model originally projected. If we considered this data back then, we might have been able to foreshadow a regression and temper expectations accordingly.

Stimson’s passing model also includes a stat that quantifies the percentage of on-ice shots that a player contributes to. This tells us how frequently offence flows through a player, which can work in conjunction with their individual point percentage to give us an idea of whether a defenceman’s point totals are misleading.

In summation, here are the tools and factors I take into account when analyzing a defencemen’s offensive profile.

  • Primary Points
  • Expected Primary Points Per Hour
  • Primary Shot Contributions
  • Defensive zone exits
  • Offensive zone entries
  • SCB %(on-ice shot contribution percentage)
  • Individual Point %(Percentage of on-ice goals a player gets points on)
  • Deployment(Teammate and competition quality, zone starts, special teams opportunities)

Case Study: Troy Stecher and Alex Edler

A 13 point decline compared to his rookie season left many fans disappointed with Troy Stecher’s offensive play.

Peel back the layers of his offensive performance though, and you’ll come to realize that his lacklustre production was the product of deployment changes and poor luck rather than inferior individual play.

The first factor to consider is that Stecher spent less than 20 minutes on the man advantage this season after leading Canucks’ defencemen in power-play time-on-ice in the season prior. By that virtue, it’d be fair to start by isolating even-strength performance for both seasons.

There’s still a notable discrepancy – 16 points in 2016-17 compared to just nine in the most recent season – but if we take a look into some of the underlying numbers we can immediately point to luck as one cause of this difference.

Ryan Stimson’s tracked data suggests that Stecher was more involved with the generation of on-ice shots and thus the flow of offence compared to last season. Despite that, he saw his individual point percentage(the percentage of on-ice goals a player gets points on) plummet compared to both last season and the 30% league average for defencemen.

Perhaps most surprising is that every offensive evaluation tool I value pointed to this season as a tangible step forward for Stecher.

 

The 24-year-old’s expected primary points rate for last season had him producing at a first pairing rate, if only barely. A big reason for this is because of the drastic improvement in his transition game – Stecher ranking among the 94th percentile in the NHL for shots and shot assists generated from plays that began in the defensive or neutral zone(Trans/60).

Such proficiency is equally impressive by Corey Sznajder’s neutral zone data, which ranked the Richmond, BC native as the third-best NHL defencemen at leading offensive zone entries with possession.

Given that strong transition play increases shot rates, it shouldn’t come as a huge surprise to hear that Troy Stecher was top-10 in the league for primary shot contributions as well. In fact, he was one of nine defenders to finish in the top-30 for shot contribution and entry rates alike.

Data courtesy Corey Sznajder and Ryan Stimson

Points, of course, still hold value. Stecher’s underwhelming counting stats speak towards his poor wrister and mediocre decision making upon entering the zone, though his rate of entries and shot contributions still created an impact that probably wasn’t reflected in his point totals.

Contrast his underlying playmaking profile with blueline scoring leader Alex Edler and you’re left wondering if raw point totals are telling us the whole story with regards to offensive play.

Data courtesy Corey Sznajder, viz by CJ Turtoro

The comparison becomes even more interesting when considering that the Canucks generated goals and scoring chances alike at higher rates with Stecher on the ice than Edler; albeit by a small margin.

Data courtesy Natural Stattrick and Corsica

Contextual differences(teammate and competition quality) often skew on-ice results, but they’re more relevant in this scenario as the pair shared four of their five most common teammates up front, while also both receiving competition commensurate with first pairing defencemen.

                   

Putting all this information together creates a conflicting picture that leaves you with more questions than it does answers. Despite a stark difference in points, not only did Stecher drive plays conducive for goals at superior rates than Edler, but the team generated scoring chances and actual goals at higher rates with the 24-year-old on the ice.

A situation like this coupled with Yost’s research on points from defencemen leaves me wondering if the offensive disparity between Edler and Stecher is truly as large as the public perception.

Conclusion

The point of this article isn’t to defend Stecher or undermine Edler’s raw production – the data tracked by Sznajder and Stimson isn’t large enough to draw definitive conclusions for this situation, even if microstats normalize after 15-20 games worth of tracking.

Instead, my objective is to challenge the simplistic method of evaluating defencemen solely based on points. Boxcar stats are always going to be a convenient and effective measuring stick to use, but we’d be doing ourselves a disservice by not also using the public tools and metrics at our disposal.

  • Interesting. I’ve always thought a D man’s points, while obviously important, like you said….don’t tell the whole tale.

    Take a playing at his peak Tanev for example…pretty much an offensive black hole, but his takeaways and transitions seemed to me, based on the old eye test, to at least create opportunity up the ice for his forwards.

    Much of the prime Hamhuis, Beiksa Edler years seemed to me to be like that as well. Such a good group of D but their point production wouldn’t blow you away. They just did their job and got the pucks efficiently to their forwards who would do the damage.

    Having said that, hopefully Hughes turns out to be what the hype is saying and we finally get a high scoring D man, cause let’s face it, that would be awesomely fun.

    • Possession has been highly under rated

      You can’t score in your own zone. This is common knowledge but it’s interesting to see it broken down and dissected statistically. I have always liked Stetcher and thought he is solid. Its unfortunate that he doesn’t have a touch more speed. If he did though, he would have been drafted etc. Unfortunately D of Stetcher and Pouliot’s size and speed have a ceiling. The exciting thing about Hughes is that he doesn’t just have good skating, it’s elite. His hockey sense is also extremely rare.

      It will be interesting to see how the Canucks deal with RHD in the coming years. There’s no urgency this season but in a season or two they are going to have a competitive forward group. At that point they are going to have to support them with a respectable D core. I like the series on the future roster but I see the most likely scenario is the Canucks go after a RHD outside the organization at some point. Relying on Jett Woo and Tryamkin is a long shot. Hopefully they don’t pull a Chiarelli though

      • I suppose it depends on what you mean by possession though. Because by “corsi” type measurements, I’d say it’s highly over rated. Those don’t correlate to wining or point totals at all, over the course of a season. The Capitals had worse 5 on 5 possession stats than the canucks last season for example.

        You’re right that you can’t score in your own zone. I think what ends up being more important is how you control the puck, and I don’t just mean when you have it. A solid D core or D man, again like Tanev for example, controls where that offensive player takes the puck. So even if we don’t have the puck, it doesn’t matter. Hamhuis was amazing at that too. Just put your opponent into low percentage areas and eventually gain the puck back. Combine that with amazing outlet passes and I think that’s what this argument boils down to. Those quick outlets and turn around plays are what lead to offense even if they don’t show up on the score sheet.

        They may have to go after another D man but I wouldn’t expect it to be cheap. There is no way, at this point anyway, that we’d be able to get a top 4 D man for anything less that a first line winger or a solid number 2 center. D just costs a lot. I think at that point they’d be better off going the FA route and trying to grab another “Hamhuis” type situation.

        • I mean the game has moved to more of a possession style. Banging pucks off the glass and dump-ins are to be avoided. You see D in the neutral zone attacking and dumping far less and circling back to regroup more (like soccer). Protecting the puck more in the O-zone.

          All these stats are a variation of possession (entries, exits). When it comes to Stetcher specifically, the problem with his size and speed is it allows for time and space in the D-zone. Guys like Karlsson and Suban (Hughes) can close the distance and create pressure. Guys like Tryamkin can separate guys from the puck or take them out of the play. The problem with Stetcher is he isn’t particularly quick for his size so he falls a bit behind in those situations. I’m using extreme examples to illustrate the point but that’s just why I have Stetcher as a solid D with a ceiling.

          You need to create pressure (tanev) and win battles (Tanev, Edler, Tryamkin) to get the puck and avoid spending a shift in your own zone. Before the Sedins started scoring they were still effective this way. They could spend a shift in the O-zone and never get the puck to the net but they weren’t in any danger and they could draw a penalty. Stetcher can be effective in his positioning and decision making like a Hamhuis but he’s never going to mature in to a game changer. The penguins figured this out with Pouliot, they were willing to give up on a top 10 pick because even if he becomes a NHL regular, the ceiling isn’t high

  • Well done Harman. I love the subtleties of altering perspective norms and working to tinker-with and improve methodology in perceiving value of individual D. Both factually and intuitively instructive. Very interesting stuff.

  • hmm, polishing a t*rd for me i’m afraid. without an elite point producer on the backend there is no chance of competing for a cup as washington, la, chicago and pittsburgh hve shown us. even recent finalists like san jose, tampa and nashville all have at least one norris calibre d-man running the show. the canucks don’t have one and it sure ain’t gonna be woo or smurf hughes imo!

    • Justin Schultz managed 13 points in 21 games for the Penguins when they won the Cup in 2017. Their next highest scoring defenceman had 9 points in 25 games. Is this what you consider elite point production?

      • schultz was a blueline beast in the absense of Letang so yes it is, 13 points and a plus 3 is awesome, one more than Ehrhoff for the Canucks in 2011 in fact. don’t cherry pick to try and paper over the cracks pal, it just makes you look bitter and foolish.

        • That beast had 2 points in 5 games in the Finals, was 4th in ice time for defencmen on his team, and was minus 1. The 2011 Canucks are another example of a team that did quite well without an elite point producer.

      • People and their “Elite Defensemen” argument (always complimented by saying a D isn’t elite because he hasn’t won a cup…). This concept of a “true number one”- who fits this in your mind? I see Doughty, Hedman, Jones, Ekblad(?)… then a bunch of Power Play specialists. I am curious who your ‘elite’ defensemen on NAS, PIT, WAS listed above are? If Subban was on the Canucks, everyone would see him as another Elder- skilled, but too careless. WAS had Calson- now at 1 good year. In terms of “smurf” Hughes- he’s the same size at the draft as some of the key folks you reference above: Keith, Letang, Karlsson… Dang Canuck fans. (they also always assume these big names are flawless and forget that they have bad games/get schooled by a Goldobin once in a while)

  • Nice analysis. As the stats and eye test agree Stecher needs a better shot to go with his strong play.. He’s been working on the shot this summer. Can’t wait to see the results. FYI Quinn Hughes also needs to work on his shot, if he improves his shot like his brother he’ll be something else.

    • I’ve enjoyedthe story of stecher. I was keeping an eye on boeser in college and heard this kid was a FA so I did my research on him. I think stecher will be a quality 2nd or 3rd pairing RD.
      Last year was a stinker for points for him as well as he didn’t look as confident as his rookie year
      But by the last 20 games I think he showed coach green what kind of player he is. I think he gets 2nd unit pp time this year. Just who else on it is a scary thought. Our total offense is gonna be bad this year unless a miracle happens. I hope stecher has a bounce back year this year

  • Great article. I do wonder if the data you pointed out is the reason why Green seems to have started trusting Stecher more as the season went, and why his bridge deal was in the $2.3M AAV area, and not closer to Virtanen territory. Likely they saw in their own numbers something similar to the above.

    I do wonder how Green will utilize Stecher this season. I can already see him getting stable first pair TOI with Edler, but will there be more PP time. Can a case be made that he would be a superior choice (along with Pouliot) for the PP than Edler? With superior shooters on each side, his lack of a good shot does not indicate a problem, and IIRC correctly, he is way faster than Edler to prevent SH breakaways.

    • Wish they signed stetcher to 5 or 6 year deal for this very reason. The guy has a good motor and is decent, he should only get better especially once some talent finds its way to the defense. Look at the scrubs he is playing with currently.

    • If he can pass better than Edler, he should be a shoo-in for Edler’s spot on the PP. Edler didn’t want to pass to Boeser and when he did, it was never in his wheel house. By the time Boeser settled the puck in his shooting position, the goalie had time to get across already. I think Stecher is a bit ‘brighter” than Edler as well, smart enough to put the puck off net or into the corner rather than beating it into the opposing forward’s shin pads and giving up breakaways. Looking forward to Pettersson on the 1/2 wall, if he can thread that cross ice pass in the NHL as he has shown that he can in Sweden, that bodes well for both the PP and Boeser. And farther in the future, Hughes CAN make those passes. I’d say looking good.

      • Up to Green to recognize this. Of Boeser’s 29 goals, Edler assisted 7 times (3 primary, 4 secondary), whereas his rate w/ Sedins was much higher. There was for certain some bias that he could not break through. Not sure how it will work with him now that it will be Pettersson there.

  • Good analysis but anyone that understands the game has never relied just on points for an evaluation of a defenceman . As myself and a whole bunch of commenters here that actually understand hockey have said (over and over and over again), analytics take second fiddle to the eye test and intangibles for defencemen. People like a few here (and one recently departed) who don’t really understand the game, can blabber on but there are facts and there are opinions. The fact is that analytics for defencemen are the third most important thing. It is more important for other position players. Driving the play out of your zone, supressing shots and goals and being a strong force to be reckoned with in the corners and in front of the net are what matter – however you do it. Defencemen need to be a thorn, not a flower (with the rarest of generational exceptions). Corwick and Fensi that however you want.

    • How do you balance that massive head on those shoulders?

      You know that “driving play out of your zone” and “suppressing shots and goals” are quantifiable and are actually measured? Zone entries and exits, Corsi/Fenwick Against, high danger scoring chances against, and expect goals against – these are the numbers that tell you if a defenceman is doing a good job breaking up zone entries, getting the puck out of his zone, and suppressing shots and scoring chances against.

      • Yes – these things are important.
        Yes – counting them is important.
        Basing every single opinion and cherry picking stats only when it suits the ANALytics and pretending the most physically demanding game on the planet can be nerdified to statistical analysis is ridiculous. IT IS A TOOL. A VERY SMALL TOOL (don’t take that too personally) especially for defencemen. The former holder of your position went out of his way to prove to us he had no idea about the realities of “the game of hockey” outside of ‘the stats’. Are you going to do the same? Freud that up….

    • There are two types of analysis in science: that which is quantifiable and can be objectively measured and that which is qualitative and requires detailed description to ensure it is both valid and reliable. Extensive experience observing something and interpreting it without consistent replicable standards still makes it wrong and worthless. That’s how we end up with CELS and armchair hickey analysts.

      The problem with many hockey stats is they do rely on observation of events that are then counted and assumed to be valid and reliable measures.

      The eye test is no better.

      A simple fact of science is that in order for any measure/observation to be valid, it must be reliable. But just because something is reliable doesn’t mean it is valid.

      Let’s no forget that science has demonstrated that one of the least reliable ways of collecting data about any event is human eye witnesses. Wrong day, wrong time, wrong person, wrong sequence of events and the human memory is exceptionally adept at changing facts to fit biases.

      Hockey is a great sport so enjoy it, because at any level, it’s pretty lousy science.

      • IMO the biggest benefit from analytics is assessing players from outside your org. Next is validating or dispelling your assumptions within the organization

        You can’t watch thousands of games each year. The advanced stats help you assess players you have limited time to evaluate. You can see how these players stack up vs each other. If you know the game, you have a clear understanding of the players you watch on a daily basis. You may have some bias or blind spots that can be checked via advanced stats also.

        The advanced stats themselves do provide info you wouldn’t pick up over the course of the season but IMO that is secondary

  • Great article, this is what we want to see on this site. Defence is the hardest position to play in the NHL – they seem to get blamed for a lot of defensive lapses yet, at the same time, they are expected to be high point producers. It’s a team game where the effectiveness of a D man is correlated to the effectiveness of the forwards playing their positions (and winning puck battles) in their own zone. The Canucks, like most teams, move the puck up the boards which makes them susceptible to pinching D from the other side. I have frequently been frustrated by the Canucks inability to get the puck out of their own zone but that is a team fault and shouldn’t be blamed entirely on the D.

  • Nice article and confirms what I see at the NHL level, and even in my own beer league, as I consider both my own defense, and watch my fellow defenders on both teams. Points are a limited measure of a defenders ability to improve his team’s scoring opportunities. Thanks for the analytics…