In case you missed it, Darcy McLeod and G-Money have developed WoodMoney, a new metric to evaluate how players have performed against different levels of competition. What they did was take every NHL player and classify them, using some pretty sound criteria (i.e., point production, ice time, Corsi Rel), into one of three categories: Elite, which are the top end players, Gritensity, which are lower end, replaceable-type players, and Middle, which are those that did not fall under the Elite or Gritensity group based on Darcy's and G-Money's criteria.
From there, they used the data available from NHL game sheets to determine how much time each player played against the different categories, and also how they performed when it came to possession and shot quality. The methodology and process to classify the players can be found at Because Oilers along with a link to the complete data-set. I highly recommend reading Darcy's article to understand the duo's rationale behind the new metric and why quality of competition is important. A couple links that explain G-Money's Dangerous Fenwick metric, which measures shot quality, is also below.
- WoodMoney: A new quality of competition metric to analyze NHL data - Because Oilers (2016, July 19)
- Explaining Dangerous Fenwick - Oilers Nerd Alert (2015, October 30)
- A Brief Statistical Look at Dangerous Fenwick - Oilers Nerd Alert (2016, July 17)
Watching Oiler games and reviewing the deployment data available on great sites like Natural Stat Trick and Hockey Stats, we know that coaches try to find the match-ups they want and do everything they can to get specific players out against specific lines. Game-to-game, we can get a sense of which players are drawing the toughest opponents, which are getting time against the fourth liners, and how they perform. And with WoodMoney, we can start to aggregate each player's outputs (i.e., shots, shot quality) against the different levels to add another layer of information to our analysis.
Two areas in particular that have been of interest to me since last summer when it comes to quality of competition are defencemen and bottom six forwards. In my opinion, neither area was addressed properly last summer by the Oilers, and it showed throughout the regular season. The team did add two experienced players in Lauri Korpikoski and Mark Letestu to their bottom six, but neither of them had a history of driving play or producing goals. As for the defence, the club added Andrej Sekera, who filled in admirably as a top pairing defenceman, and saw their faith in Brandon Davidson payoff very nicely. But the Oilers took a risk, an unnecessary one, starting the season with Griffin Reinhart, Eric Gryba and Andrew Ference.
Because of these decisions, it was worth following how the coaching staff was going to deploy defencemen and depth forwards in the hopes of winning games this past season. Using the WoodMoney data, we can start to dig into this a little more and try to uncover any patterns in player usage and what the Oilers could potentially build on going forward.
Before moving ahead, a quick glance at how the categories are defined when assessing how each player did against different levels of competition.
- Elite: at least one elite player must be on the ice. No Gritensity players may be on the ice.
- Middle: Elite + Gritensity on the ice or 3 muddles.
- Gritensity: Any time a Gritensity player is on the ice except when with an Elite (Source: Because Oilers)
Please note that WoodMoney is still being refined based on further data analysis and testing, so you may see these numbers, and possibly the rationale for categories, change over time. It's a new metric, one that has been very transparent when it comes to methodology and raw data and remains open to feedback.
First up, a graph of every Oilers defenceman and their proportion of ice time at 5v5 against the three categories of forwards this past season. All three graphs are sorted by the percentage of ice time against the Elite category.
Here we see Klefbom, Sekera, Fayne and Davidson seeing the Elite players for 35% or more of their total ice time at 5v5. This checks out as we know these four played top pairing minutes for extended periods of time. Davidson played primarily bottom pairing minutes, but he did have a stretch of games playing top pairing minutes right before he got injured. Klefbom was relied on heavily by the coaching staff over his 30 games to take on Elite competition, and as we see below, it was for good reason.
Here's how the Oilers defence core did when it came to Corsi For% against the different categories of forwards. Again, it's sorted by the players proportion of ice time against the Elite's, so we have the top pairing defencemen listed first.
Here we see that that although Klefbom and Sekera got the most minutes against the Elite category, it was actually young Davidson who performed the best when it came to shot shares. With a Corsi For% just under 50%, the Oilers did a lot better against the Elite's with Davidson on the ice (CF% Rel of +2.10). Klefbom and Pardy were the only other defencemen with positive relative stats here, with the rest falling below the team average.
Finally, I graphed out how each player did when it came to Dangerous Fenwick, which gives a weighting to each shot to predict it's chances of being a goal. It measures shot quality similar to the Corsica Hockey's expected goals metric.
Here we see again Klefbom and Davidson playing relatively well against the Elite category, as the team did better at getting their share of quality shot attempts with them on the ice. The rest of the team lags behind, with Fayne holding the worst DFF% Rel against Elite's with -3.5%. So while he did alright when it came to shot attempts against the Elite category, the team allowed more quality chances against when he was on the ice.
Update: I've also put together the Corsi Against/60 and Dangerous Fenwick Against/60 rates for each of the defencemen (thanks to Perry K for the friendly reminder).
Davidson still looks great using the rate of shot attempts and dangerous Fenwicks against, especially against the Elite grouping, as does Klefbom. The other young defencemen appear to have done alright against the Elites, but Reinhart especially appears to have struggled with the Middle category and lower level players. Gryba also looks good here, so I'd be curious to see a split of how he did with and without Davidson as his partner.
Oilers Bottom Six
Below are the most common bottom six forwards from this past season, along with their proportion of ice time against the three different competition bands. Keep in mind that from this group Matt Hendricks, Mark Letestu, Lauri Korpikoski and Iiro Pakarinen fell under the Gritensity category largely based on their poor production and rel-stats.
Here we see that Yakupov and Kassian, who I decided to include here, saw more ice time against the Elite category, which makes sense as they both played extended stretches in the top six. Yakupov did have a nice run with McDavid, but his most common centerman this past season was actually Letestu by about 30 minutes (Source: Hockey Analysis). As for Kassian, he did get to play next to Hall and Draisaitl for a stretch, but he finished the season on the fourth line with Hendricks and Letestu. To put things into context, McDavid, Hall and Eberle played between 35-40% of their ice time at 5v5 against the Elite category. Going forward, the Oilers should ideally be acquiring depth players that can be relied on to play against scoring lines, so hopefully that will be addressed this summer.
Next, we see how the bottom six forwards did when it came to Corsi For% against the different categories. Since they spent most of their time against the Middle and Gritensity category, you would hope that they put up respectable numbers in those match-ups.
Based on the data, it appears that against the Gritensity group, the Oilers depth players barely hovered around the 50% mark, with Anton Lander posting the best numbers among the group. Against the Middle category, Kassian and Yakupov posted the best numbers, but that may have been driven by their ice time with top line players. Among the regular bottom six forwards, there's little to highlight. This group got absolutely owned when it came to the proportion of shot attempts, which has got to be better going forward. The WoodMoney metric confirms what we saw a lot of this past season: a bottom six that couldn't match up against depth players, putting more pressure on the top six guys to produce.
And finally, here's how the depth forwards did when it came to Dangerous Fenwick. It gets uglier.
So while the depth guys did alright against the Gritensity category when it came to shot attempts, they actually posted dreadful numbers when it came to the quality of shots, allowing more dangerous shot attempts. Remember, the Gritensity category are the low-end, replaceable players with low ice time and poor production numbers. But against the Oilers depth players, they would often win the share of total quality chances. Against the Middle category, only one of the regular depth forwards had a DFF% over 45%.
We knew throughout the season that the Oilers had poorly constructed their defence core, and that there were some depth issue up front. But seeing how exactly they did against the different levels of competition gives us a sense of how much trust the coach had in the players and what areas need improvement heading into 2016/17.
I appreciate the time and effort that Darcy and G-Money have put into this new metric and largely agree with their methodology. Again, this is new territory when it comes to quality of competition so it's worth validating what we saw this past season and how the metric aligns with existing concepts.
Curious to hear what the C&B community thinks of WoodMoney and my findings above.