clock menu more-arrow no yes

Filed under:

Taking Center Stage: An analytical comparison of Matt Duchene and Music City’s centers

New, comments

As the rumors swirl, the “childhood team” stories start, and the market heats up, is Duchene worth the heartache? Or, more importantly, the price tag?

NHL: All Star Game-Skills Competition
Jan 30, 2016; Nashville, TN, USA; Central Division forward Matt Duchene (9) of the Colorado Avalanche skates with a cowboy hat during the breakaway challenge during the 2016 NHL All Star Game Skills Competition at Bridgestone Arena.
Christopher Hanewinckel-USA TODAY Sports

By now, it seems all but inevitable that Matt Duchene will finally make his long-awaited arrival in Nashville when the Free Agency period opens on July 1st. Despite rumors that Duchene’s contract could command over $9.5 million AAV, Predators GM David Poile seems set on making this addition—especially following the heartbreaking draft-day trade of P.K. Subban.

I wanted to compare the play of Matt Duchene to the centers that are currently on the Nashville roster. For this exercise, there are a few things to note:

  1. I used data from Evolving Wild/Evolving Hockey from the past two years to get an idea of recent performance.
  2. The exception is the tracking data compiled by Corey Sznajder and visualized by C.J. Turtoro. This data makes use of hand-tracked statistics, and, to have a decent sample size of games tracked (as this is a monumental effort), those numbers come from the prior 3 seasons, about an average of 20 games per season.
  3. Nashville’s “centers” for this are defined as Ryan Johansen, Kyle Turris, Nick Bonino, and Brian Boyle. Boyle spent some time centering the fourth line towards the end of the season, and has a larger sample size of games than the other Nashville center, Frédérick Gaudreau (and if you know me, I am loath to exclude Freddy G from anything).

With those assumptions in mind, let’s jump into the data. If you’d like to delve into the visualizations yourself, you can find them on my Tableau page.

The Basics: Individual Statistics at Even Strength

Individual Statistics at Even Strength - 2017-2019.

It is immediately apparent that Duchene brings a consistent level of production across the board in the more basic stats. He leads every category by a fairly decent margin with the exception of shooting percentage, where he finishes only a hair behind Nick Bonino (surprisingly). Duchene is producing points, from both goals and primary assists, at a rate unmatched by any Predators center—and he’s doing it with high quality shots, as evidenced by his 0.78 expected goals (xG) rate.

His shooting prowess is further on display when comparing his xG/Goal differential of +0.24—he’s scoring better than his quality would suggest. Bonino (+0.24 as well) is out-shooting his level of quality as well, but producing at 75% of the rate of Duchene. Depending on your opinion, this could also suggest that one or both of these players are due to regress to the mean from their stellar shooting performance, but based on utilization, Bonino is the more likely option for regression.

The low goal-scoring rate for Ryan Johansen and Kyle Turris seems like it might be a warning. However, the underlying stats suggest Johansen is performing at a high level, albeit in a different way. Johansen may trail Duchene by 0.41 goals/60, but he’s producing primary assists and points at a level in Duchene’s range (0.84 vs 0.77 A1/60 and 2.28 vs 1.99 P/60). His shooting percentage, at 11.7%, is not bad—in fact, it further cements a thought that every Predators fan who watches Johansen has had over the last two seasons: Ryan, just shoot the puck. He has the ability to score when he does, but he serves primarily as a facilitator for his elite linemates in the JOFA line.

Kyle Turris has a similar issue; however, a dismal 6.8% shooting percentage highlights the trouble he and the second line as a whole have experienced for the last 14-15 months.

The Sore Spot: Individual Statistics on the Power Play

Individual Statistics on the Power Play - 2017-2019.

I’m sorry I had to remind you of the power play.

Duchene and Johansen received the bulk of the minutes on their respective teams’ power plays; however, the main difference remains goal scoring. This time, though, Johansen’s shooting percentage has completely tanked at a meager 2%. Bonino and Boyle both outscore Johansen in terms of pure goals, while Kyle Turris outperforms everyone, despite receiving less minutes than the first-line centers. He presents an interesting option for the power play unit going forward (if he remains on the team going forward).

Of the group, Bonino is the only player who does score goals but without driving the offense in any meaningful way. If the Nashville Predators move to utilizing 4F1D for both units, this group of centers provide some options. But the story remains the same—Duchene would again be an upgrade to the team. Hopefully new power play coach Dan Lambert will also be an upgrade to the team.

Teamwork: On-Ice Statistics at Even Strength

On-Ice Statistics at Even Strength - 2017-2019

The on-ice results reflect the performance of the team while the player is—you guessed it—on the ice, and is a good barometer of the team surrounding the player. Duchene wouldn’t find reliable teammates until his recent stint in Columbus, being anchored down to awful teams in Colorado and Ottawa. Still, the usefulness of differential statistics is somewhat limited, and are best used to compliment other analysis to get a better picture of the entire team.

The Facilitator: Tracking Data at Even Strength

Tracking Data at Even Strength - 2016-2019. Data courtesy Corey Sznajder and CJ Turtoro.

As much of hockey Twitter will tell you, there's only so much you can do to analyze the game by putting play-by-play data into a spreadsheet. And, while that is the preferred method for numbers geeks such as myself, only by manually tracking games can you get the extra data used by the teams and, soon, the NHL itself.

For three years, Corey Sznajder has hand-tracked around 15-35 games per team a season, counting everything from zone entries and exits, dump-ins, and passes in the offensive zone. And from this data, we can get a new level of insight—and we are eternally grateful for his work (go support him on Patreon here if you are able!).

Folks, Matt Duchene has been at or near the top in every metric we've examined so far, but this time, he has company: our beloved 1C, Ryan Johansen. Since his arrival, Johansen’s detractors have pointed to his dwindling point totals as proof the trade was bad. But, looking at this data, one can see that he is absolutely elite at generating offense when he's on the ice.

He resides in the 98th percentile of the league in shot assists per 60—this differs from game data, as tracking data allows for counting of tertiary assists. When counting shot contributions (shots + assists / 60), he's still in the 87th percentile of the league, despite being in the bottom 12 percent in shot volume. With dangerous goal scorers like Filip Forsberg and shot-volume master Viktor Arvidsson, why not play to your strengths and create offense so those guys can score? They're only numbers 1 and 2 in team history for single-season goals.

Duchene is very good in his own right, albeit a bit more balanced across the board, and this could work to Nashville's advantage. Where Johansen can use elite passing to set up his wingers and rely on them to enter the zone, Duchene is capable of not only carrying the puck down the ice, but also scoring, setting up high-volume Craig Smith, or passing to Mikael Granlund, who is no slouch in generating offense himself. He brings an extra kick in overall offense to the second line that Turris, while being good in his own right, lacks. These two would be the centerpieces of a truly dangerous top six.

The Model: Wins and Goals Above Replacement

Wins and Goals Above Replacement - 2017-2019

Last is a look at Evolving Wild's model for Goals Above Replacement and therefore Wins Above Replacement. This advanced statistical model evaluates skaters to establish how much each player contributes to the team’s goals and wins—and how much better (or worse) they would be than a "replacement" level player (the average performance of a player that would be called up in case of injury, etc.).

Overall GAR is comprised of components: PP offense, SH defense, Penalty GAR (which is another long discussion) and EV GAR—consisting of both EV offense and EV defense. For offensive components, adjusted Goals For are used: a player contributes most to wins by scoring goals when on the ice, and some players score a lot despite low quality of shots (think Patrik Laine). However, on defense, using goals against has issues: so many factors go into what makes a shot into a goal—what kind of pre-shot movement is there, how good is the goalie, did the shot get a lucky bounce, etc. But a player can contribute to defense by limiting the opponent’s shot quality; playing sound defense force the other team to take lower quality shots, and a skater has more control over that. So for defensive components, the rate-adjusted expected goals against (xGA) is used.

The good news? Duchene is again in a class of his own in comparison. The bad? While Duchene leads by a lot in overall GAR, he and Bonino share an identical EV GAR value, while Johansen sports a meager 0.096. For most Predators fans, this seems counter-intuitive. But in terms of the model, it makes some sense: Duchene is elite in rate adjusted Goals For—he’s a scorer—but he’s average at defense (xGA). Bonino, on the other hand, is very good at limiting shot quality, but average at on-ice GF; we all know his line isn’t really known for scoring prowess.

Johansen, however, is exceptional at offensive GF (the JOFA line is the scoring line, after all), but he also contributes a very below-average defensive component because he allows a lot of quality shots against. Even though he sports a positive goal differential while on the ice, he’s often facing opposing top lines, and he (and the team as a whole) is saved by Pekka Rinne more often than not stopping a large volume of high-quality shots.

For the reasons stated above, there are arguments for and against this and other WAR models, but I see it as a very reliable tool in the toolbox of analyzing players. The "eye test", basic and advanced stats, and statistical models are all tools available, that, when used together, allow for a more complete evaluation of a player in the NHL today. And even if you feel the flaws of this model are too much to overlook, it is still telling that Duchene performs the best overall out of the entire group—he’s an addition to this lineup no matter how you look at it.

Conclusions

Matt Duchene is a great top-six center and would be a good compliment to Granlund and Smith on the second line. Is he worth it, especially if the asking price is north of $9.5 million AAV? I’m not sure.

This would limit the ability to sign basically anyone else, since Grimaldi and Sissons seem to be locks to re-sign as RFAs. But at this point in time, the damage has been done—Subban is gone, with only two second round draft picks and prospects to show for it so far. If Duchene does not sign with Nashville now, it will be far worse an outcome than overpaying to get him.

One byproduct of this study was that I found myself wanting to give Turris a chance on the team alongside Grimaldi and Boyle, with Bonino remaining on the Watson-Bonino-Sissons line that the coaching staff seemed intent on keeping together. Poile has played his hand—somewhat obviously—and teams are going to be unwilling to help unload contracts such as those of Turris and Bonino, which most believe should be necessary. If the centers on this team have to consist of Johansen, Duchene, Turris and Bonino, it would be an improvement from last year. It’s not the saving grace the team may think it to be, but Duchene is too talented to not have an positive impact.

Let’s go for it: nothing can be worse than losing Subban for scraps, right?


Data courtesy Evolving Wild (@evolvingwild), Corey Sznajder (@shutdownline) and CJ Turtoro (@CJTDevil).