Matchups matter in hockey. We see coaches send out certain lines and pairings to go up against someone on the opposition in games. We hear about how having the last line change can be a big deal tactically in a game - and it can be. We understand that a matchup that has gone well or gone awry can make a difference in how a game plays out. From a intuitive standpoint, I think most fans would recognize the difference between succeeding against most first lines in the league and a fourth line. Success is good, but one is more impressive than the other.
From a statistical standpoint, there have been attempts to quantify quality of competition. A player’s average time of ice of their competitors have become one of the more common ways to define this with a number. There have been other attempts, using the opposition’s raw or relative Corsi when the player is on the ice. However, neither way really matches up with reality. We know Adam LArsson and Andy Greene get a lot of tough matchups, but it wasn’t as if they had nothing but. Likewise, we know Eric Gelinas was protected but he that was not always the case. Those metrics miss that. There’s a newer and more accurate way to define it that I think should be recognized that does. It’s the WoodMoney metric for quality of competition.
The What Metric?
It’s called WoodMoney as it was the collaborative effort of Woodguy55 of Because Oilers and G Money of Oilers Nerd Alert. They made it so they named it and I can respect that. It’s also a lot better than G-Guy55. Anyway, Woodguy55 has a full explanation of how it was devised and how it works at Because Oilers.
The long and short of it as I understand it is that rather than lumping every matchup a player faces together, quality of competition is split up into three bands: “elite” forwards, “dregs” forwards, and “muddle” forwards. The elite forwards are the best of the best and only lines full of them are elite. The dregs include an array of players buried in depth charts, call-ups, fringe players, and so forth - and those players. A line is a “dreg” line if they’re all dregs. The “muddle” is everyone else - including lines that feature an elite and a dreg on it. A player’s time on ice against those three categories is then established. Within those sub-sets of data, a player’s relative Corsi and Dangerous Fenwick (more on that in a bit) is calculated for both in comparison to their teammates against all levels of competition and within that level of competition. Essentially, the WoodMoney metric provides more specific quality of competition in 5-on-5 play than just the on-ice opponent’s time on ice or Corsi alone. It recognizes the reality that not every match-up is set and players will be forced against all levels at some point in a season.
The available .csv dataset at Because Oilers has this metric for all 2015-16 players against Eastern and Western Conference opponents, home and away splits (which should be very interesting to breakdow), both (which is what I’ll be using), and it includes a new metric I haven’t seen before along with Corsi: Dangerous Fenwick. This is something G Money has figured out and has explained what it is and why it matters at his blog, Oilers Nerd Alert. It’s similar to Emmanual Perry’s Expected Goals stat. One of the limitations of Corsi (all shooting attempts) and Fenwick (unblocked shooting attempt) is that all attempts are counted equally. However, we know that a shot from the slot is more desirable - or dangerous - than a wrister from the blueline. After taking goal location and shot types for five years, G Money found that the average dangerous unblocked shooting attempt is a wrister from 29 feet away. G Money weighs that as 1 dangerous Fenwick and bases that for other shots. A shot that is closer is a higher value, a shot that is further away or not as threatening is not. For a defenseman, this can identify whether they are on the ice for more threatening shots than not - which tells us more than just CA/60 or SA/60. I think that’s worth pointing out in conjunction with the WoodMoney metric for quality of competition. Plus, it’s already there.
So I think this is all interesting. And when I usually find something interesting regarding stats in hockey, I immediately think: how does this apply to the New Jersey Devils? (And, can I understand this enough to write about that?) Woodguy presented the metric with a defenseman in mind - former Devil Mark Fayne. On top of that, Gerard recently wrote about potential teammates on each line and pairing, C.J. wrote about what the numbers say about Jon Merrill and John Moore, and Alex touched on the relative youth of the defense. Most of all, the New Jersey Devils traded Adam Larsson for Taylor Hall and that leaves a significant role left to fill on the blueline. By using the WoodMoney metric, we can look at how Larsson performed against the varying levels of competition, how the other Devils defensemen did, and use that information to make a more informed guess on who can take that spot.
The 2015-16 Devils Defensemen by the WoodMoney Competition Band
There were twelve defensemen who suited up for New Jersey for at least one game last season. However, I’ve focused on the defensemen who played at least 200 minutes of 5-on-5 play last season. This was enough time to ensure everyone got at least a taste of each level of competition. Plus, I don’t think there is much to gain from the seven games of Vojtech Mozik, the four games of Marc-Andre Gragnani, and the one game of Steve Santini. That said, here’s the breakdown by competition for the other nine defensemen:
This immediately passed the smell test for me. Greene and Larsson took on the toughs more often than not - and they did per this graph. The group of Damon Severson, John Moore, David Schlemko, and Jon Merill were mixed between the second and third pairings so they had a good amount of time against the muddle and the dregs. Gelinas’ protection can be seen given his high percentage of ice time against the lesser forwards of the league. Surprisingly, Seth Helgeson’s call up was not as protected. Unsurprisingly, David Warsofsky’s breakdown was more like Gelinas’s breakdown. This all establishes how the defensemen were used on New Jersey last season. Now let’s see how each did against each level of competition.
The Top Pairing: Greene and Larsson
Let’s start with Greene. Greene has been very good for the Devils. He’s grown into the role since Paul Martin left New Jersey and has done a very good job. Unfortunately, the 2015-16 Devils got rolled in possession. Corsica has the team’s CF% at 46.17%, the second lowest in the entire NHL last season. And do recall that the Devils have been a low event team for the better part of the last nine seasons. What this means is that these possession numbers are going ugly:
(Note: For spacing purposes, I shortened CompTOI% to C-TOI% and CF%RelComp to CF%RelC. The data still was pulled all the same from Woodguy’s and G Money’s spreadsheet.)’
Greene did play quite a lot against the elite - and got rolled himself. His CF% against lesser players was better, but still under the breakeven point on the team. What’s concerning is that other defensemen performed better from a possession standpoint when Greene got off the ice within that level of competition. That’s what the CF%RelC means - the CF%RelAll is compared against all levels of competition. The latter’s values show that Greene really suffered against the elite but was a positive force against lesser competition. But within only that level of competition, he was not.
I’ve included his CF/60 (Corsi For per 60 minutes) and CA/60 (Corsi Against per 60 minutes) to really show the root issue here. For the better part of the last nine seasons, the Devils as a team has been fantastic at limiting attempts against and fantastically awful at generating attempts. Greene’s 53.1 CA/60 against elite competition is actually not that bad within the WoodMoney dataset. It’s that terribly low CF/60 that leads to a hideously low 38.9% CF% against elite competition. What does Dangerous Fenwick show? Let’s see:
In terms of Dangerous Fenwick (DFF), the same issues with percentages apply with respect to elite competition. Other Devils defensemen have been better with respect to Dangerous Fenwick for and against the team than Greene, whether its within elite competition or not. Fortunately for Greene, he looks much better against lesser competition. As with Corsi, the issue isn’t so much Greene has been present for a lot of dangerous shots. It’s that the Devils haven’t created many of their own when he’s out there.
To an extent, this is both a problem for Greene and the Devils itself. This shows that Greene’s not pushing the play forward. While the position is called defenseman, its not really good to spend most of the shifts stuck in your own end of the rink, just trying to ward off the opposition. That said, Greene’s teammates certainly haven’t created much either and I suspect until they get better, Greene won’t be better for it.
That’s a lot about Greene. I started with him because these numbers show similar issues for Adam Larsson.
We can see from these charts that Larsson was a bit better than Greene last season. They were about the same against the elite; they both struggled from a possession standpoint. A CF% of 38.8 is just, well, a problem. Larsson was a bit better with respect to Dangerous Fenwick and CA/60 against the Elite forwards, though. Larsson had better percentages against the Muddle and the Dreg forwards of the league than Greene. This led to Larsson having better percentages overall compared with Greene. However, these numbers suggest that perhaps this hole is not as big as it may seem. While they don’t include the real factors of the minutes (Larsson averaged 22:30 last season in total ice time) and situations involved (e.g. lots of defensive zone starts, lots of penalty killing), Larsson still suffered like Greene against the best of the best forwards of the league. Larsson is quite good and Woodguy himself argued as such in an earlier post at Because Oilers; but these tables tell me that Larsson (and Greene) weren’t the shutdown defenders against the top competition as used.
The Remaining Defensemen: Severson, Moore, Merrill, Helgeson
I’ve got a table for each defenseman. I’ll be more brief in my commentary for each given that I think it’s clear what to look for with each one given what I wrote about Greene and Larsson.
Severson has been touted as a potential replacement for Larsson’s role. While Severson has not played a lot against the Elite, his relative CF% relative to his teammates within that level of competition is actually positive. It’s also positive for relative Dangerous Fenwick percentage within that level of competition. Both are encouraging signs that he could do it. His CA/60 and DFF/60 at that level, though, are not as encouraging. They are quite high, especially his CA/60 against the Elite. But the Devils may have to deal with giving up a few more attempts to have someone who was. If John Hynes and the other coaches are not so comfortable with that, then Severson can continue to be used more against the non-elite competition. He wasn’t as great against the larger group in the Muddle category as he was against the lesser competition. If Larsson was still in New Jersey, I would recommend he perform better against them before using him in a bigger role. He may have to jump into the lake and see how well he can swim.
Now let’s look the more experienced John Moore.
C.J. pointed out that Moore was a liability in possession. Breaking it down by competition really shows where that is. While his CF% and DFF% in the Elite category is superior to even Greene and Larsson, his CA/60 and DFA/60 are much higher. What’s more concerning is that he was not much better within his more common level of competition, the Muddle group of forwards. Basically, Moore was used as a top four defenseman and, well, he had to play a lot of defense and a lot of attempts came Schneider’s way when that happened. That’s not that good. Against the weakest level, yeah, Moore looked much better but he wasn’t as good as Severson. And I don’t think the Devils can really afford to keep Moore against limited competition at this juncture.
Let’s move on to another defenseman C.J. recently wrote about, Jon Merrill:
More than Severson and Moore, Merrill was kept more against the Muddle and Dreg forwards. As it turned out, Merrill actually did pretty well against the Dregs. He came pretty close to breaking 50% in terms of CF% against the Dregs and he did do it with respect to dangerous shots. Against more difficult competition, well, Merrill did not do as well. While he did not play nearly as many minutes against the Elite as he did against the Muddle group, it’s odd he did better against the Elite from a CF% standpoint. From a DFF% standpoint, he didn’t - so I am not as confident of this being a positive sign for Merrill. I’m looking at this and C.J.’s post, and I’m agreeing with him that Merrill really is a third-pairing caliber player - at best and for now. Could he improve? Possibly. But there really isn’t anything to be excited about here.
I’m surprised at Seth Helgeson’s time on ice breakdown against the three bands of competition WoodMoney uses. He’s still on the team and could possibly be in the mix for being on the roster when training camp comes along. And I fully expect him to be called up if there are a number of injuries on the blueline. He did that last season; here’s how he did:
Helgeson only looks good relative to others against elite competition, which is a curious finding. But he didn’t play much against them and his rates against are just high. He was way, way worse against the Muddle band and the Dregs don’t show well from a Corsi perspective. This doesn’t really show Helgeson as anyone different than one would think.
The Other Departed Defensemen: Gelinas, Schlemko, Warsofsky
Eric Gelinas was dealt during last season; and both David Schlemko and David Warsofsky joined new teams, San Jose and Pittsburgh, respectively. While their departures were not as big as Larsson and the team has the bodies for depth, the Devils will still have to replace them to some degree To that end, let’s briefly look at how they did. Think of it of what the others, including call ups, may have to face.
Eric Gelinas’ data is likely confounded with his six games in Colorado. But he did appear in 34 games last season and he’s always been a statistical curiosity. By eye, he doesn’t appear to be a good defensemen. However...
These numbers really aren’t that bad. Gelinas was the only defenseman to come out at 50% or better in Corsi against each band of competition. While his CA/60 rates were high, there was offense when he was out there to balance it out. Hynes and his staff limited his action so there’s definitely a “sample sample size” factor against the strongest competition. While I may remember some horrid moments, overall, he wasn’t wrecked. I still think he’s best suited when he’s protected in 5-on-5 and used primarily in offensive situations given his skillset and his other issues. But the numbers show that he wasn’t a disaster last season.
David Schlemko was seen as a good hand last season for the Devils. While not as emphatic as Lee Stempniak, Schlemko’s signing was a pleasant surprise for the defense last season. It’s arguable his 2015-16 revitalized his career as it led to a four-year deal with the Sharks. Here’s how he earned it:
Again, not bad. Schlemko definitely earned more of the coaches’ trust, which has resulted in a more even usage between the three levels than Gelinas. Injuries to others also played a role in that, too. I was surprised to see that his proportion of on-ice CF% and DFF% was better against the best than the larger mid-range of forwards. That doesn’t mean Schlemko was someone they should have used more; just that when he had to do it, it wasn’t so bad relative to the other Devils. Still, this shows that Schlemko really shined against the Dregs. I don’t think the Sharks gave him four years to just do that, but with a stronger set of forwards in front of him and a possession team that isn’t coming off being the second-worst in the league, I think Schlemko can do well in San Jose.
David Warsofsky was a waiver-wire pick-up that ended up playing quite a bit in his ten appearances. Specifically, an average ice time of 16 minutes. So he made the cut; here’s how he did with respect to how he was used:
Warsofsky’s usage was similar to what Gelinas got last season. Both are offensive-minded defensemen who are seen as risks on defense, so that’s understandable. Warsofsky actually did well against the Dregs in both categories and did well against the Muddle crew of forwards in a possession standpoint. And the team’s CF/60 was actually quite high when he was out there. So he was at least present for that going well. Dangerous Fenwick, well, that was not as impressive as total attempts. It means that what was created was not so threatening as what the opposition did. Still, for ten games, Warsofsky did well against his most commonly faced competition.
Lastly - The Incoming Ben Lovejoy
In case this wasn’t exhaustive enough, let’s look at these numbers for the newest New Jersey Devil defenseman: Ben Lovejoy. While I wasn’t excited over his signing, these tables show something encouraging:
Apologies for leaving in the raw CF and CA that Lovejoy faced. But this is actually something I did not know about him. The right-handed defenseman the Devils just signed actually didn’t too bad against all three levels of competition. He clearly performed better against the non-Elite level of competition, he played on a team that wasn’t so low in CA/60, and that team was Pittsburgh so he had some support from excellent players. According to Hockey Analysis, Lovejoy’s WOWY shows that he did play over 200 minutes with Sidney Crosby, Phil Kessel, and Evgeni Malkin. So that can’t be ignored with respect to Lovejoy. That all said, he didn’t fare too badly against top level competition. Given that the Devils don’t have any other defensemen that did as well against such a level for a good amount of ice time, my takeaway from this Lovejoy could be an option as a Larsson replacement. At least from this standpoint. If the coaches go in another direction, then I am at least more encouraged that he could handle the second pairing. Possibly as someone to help Moore with his issues.
Concluding Thoughts or So What of All This?
I will admit that I threw a lot in there. A new approach to quality of competition plus a new stat with it. Both applied to ten defensemen, nine who played at least two hundred minutes for New Jersey, and one who will most likely play quite a lot for the Devils next season. So what does this all mean?
Starting at the top, Greene and Larsson were indeed used against top competition more than anyone else on the team - and they suffered from a possession and dangeous Fenwick standpoint. What it tells me that is that while they played an important part of the team’s defense and their usage was far from easy, they did not do as well as thought. Top competition fed on them as they fed on the team. When they did face lesser competition, things were better but still in the red. While I think that’s related in part to the team being in the red in terms of possession, particularly due to not generating a lot of offense, I’d be lying to say that they couldn’t have been better. For the Devils to get back to some kind of prominence, they’ll need to be better. While this metric doesn’t touch on quality of teammates, I wonder if that’s the path going forward. Again, Larsson’s and Greene’s CA/60 and DFA/60 against the elite wasn’t bad; but the CF/60 and DFF/60 was abysmal. That’s how I see it, but it looks like Greene and whoever else will take Larsson’s spot next season could stand to be better too.
Who will take Larsson’s spot? This remains as the big question for the Devils’ defense. I don’t think there will be a definitive answer. But by looking at this, I can forsee it coming down to Severson or even Ben Lovejoy. Granted, this is only one aspect of a player’s game. I don’t know how either can handle more minutes, tougher zone starts, and how well they would play with Greene and/or the forward line that they’ll play with. I’d like to favor Severson in the hopes that he will become a better player and that he actually has offensive skills. Lovejoy, not so much, but he brings experience with him and that may be preferable as a stop gap until Severson gets better or someone else emerges. Neither are ideal options, but by the WoodMoney metric, it’s was not a very ideal first pairing per the stats used within it.
If nothing else, I don’t see anyone else who was on the team last season to really emerge. Using the WoodMoney metric supports the notion that Moore and Merrill are not well suited for more minutes. It’s arguable that Schlemko will be missed to a degree, although Lovejoy or Severson could fill in nicely - perhaps even better than whoever ends up with Greene. Warsofsky and Gelinas came out better than I would have guessed. If one of Mozik, Yohann Auvitu, or Santini can be protected, then that may be a good way to see if they can become a regular in this league. One of those three would end up being depth for call-ups and, hopefully, perform better than Helgeson. I still wouldn’t be too surprised if the Devils bring in someone on a try out in camp or sign someone with an offensive skillset on defense that can be used like Warsofsky and Gelinas was used last season.
The big question on defense isn’t going to go away, and the smaller questions also remain. But at least it looks a little clearer as far as what options seemingly make sense on defense.
All the same, I think the WoodMoney metric has some legs. I hope others take a closer look at the data, whether its for their favorite teams or I also hope Woodguy and G Money are able to get their planned site up for it, or at least provide updated spreadsheets for download throughout 2016-17. I really like their approach to this to include different levels of competition, and their reasoning behind it makes sense to me. It accounts to a degree the level of how matchups are used, which is a real thing in hockey. I think the Dangerous Fenwick stat is something to keep an eye on as well, too. Between the two, I think there’s at least a little more insight to gain as to how a defenseman performs in 5-on-5 play. And it can be done for forwards as well. And, as a final wish, I hope Woodguy and/or G Money (or someone else) consider something similar for quality of teammates.
What do you think of all of this? What do you think of the WoodMoney metric for Quality of Competition? What did you learn from how it was applied for the Devils’ defensemen (and Ben Lovejoy)? Do you think it provides some insight as far as how the Devils should use their defensemen for next season? Please leave your answers and other thoughts about the metric, the stats, how the Devils defensemen did by it last season, and the defensemen for next season. Thanks to Woodguy and G Money for having the data available and coming up with the metric; thank you for reading.