There are many statistics that allow for the evaluation of hockey players. Corsi and Fenwick give an idea of which zones players are spending their ice time in and how involved they are in sending pucks toward opposing nets or getting them sent at their own. Quality stats allow a glimpse at the calibre of opposition or that of one’s teammates. Zone starts show which players are depended on to prevent goals against and which are used to score them, and also how well a team fares with the hockey equivalent of football’s field position.

All of these stats are important to understand a player’s value as one stat can be heavily influenced by another. The way this comparison is usually done is by stating statistics separately, often placing them in a table with a player ranked by position on either his team or the league. ExtraSkater.com has great graphical player usage charts
that plot many of the stats listed above.

In this article I will outline my attempt to combine these various statistics in an effort to create a means of evaluating a player’s effectiveness both in relation to his teammates as well as to his peers on other teams in the league.

For this, I chose to look at the defencemen in the NHL. I wanted to include as many of the defenders who have played for the Canadiens this season in the analysis, so I collected data on those who have played a minimum of 15 games, or 21% of the current season, to include the rookies Nathan Beaulieu and Jarred Tinordi. This created a pool of 237 defencemen who have reached that criterion in the current season.

The first phase of this evaluation factors in quality of both competitors and teammates for each defensive player, as well as how that player is zonally deployed. The result is a measure of the difficulty of each defenceman’s ice time I’ve termed Deployment Quality (DQ).

Deployment Quality (DQ)

This evaluation metric uses the statistics (from >) listed below. For each stat, ExtraSkater’s glossary definition (‘ESGD’), range of values in the obtained sample, and median of the sample are shown:

Quality of competition by average ice-time (5v5) (TotTm% QoC)

ESGD: TotTm%: Percentage of team's total ice time that player is on ice for
ESGD: TotTm% QoC: Average (time-weighted) TotTm% of 5-on-5 on-ice opponents

Range: 30.2% (Dion Phaneuf) to 26.7% (Shane O'Brien)
Median: 28.5%

Quality of teammates by average ice-time (5v5) (TotTm% QoT)

ESGD: TotTm% QoT: Average (time-weighted) TotTm% of 5-on-5 on-ice teammates

Range: 30.5% (Carl Gunnarsson) to 24.7% (Clayton Stoner)
Median: 27.2%

Defensive zone start percentage (DZSt%)

ESGD: Defensive zone start percentage at even strength

Range: 41.1% (Paul Ranger) to 22.1% (Sheldon Brookbank)
Median 31.5%

From these pilfered statistics I calculated some others (you can view all of the stats for this project in a publicly-shared spreadsheet):

Difference in quality between competition and teammates [TotTm% Qo(C-T)]

This was calculated by subtracting each player’s teammate quality (TotTm% QoT) from his competition quality (TotTm% QoC).

TotTm% Qo(C-T) = TotTm% QoC - TotTm% QoT

Range: 2.7% (Dylan Olsen) to -1.2% (Jonas Brodin)
Median 1.2%

The higher the number, the greater the disadvantage between the quality of the players a defenceman plays with and those he is put on to play against. The lowest numbers indicate an advantage in quality between teammates and competition.

Quality of On-Ice Personnel (QoIOP)

This number is obtained by taking the competition quality (TotTm% QoC) stat and adding the difference in quality between competitors and teammates (the TotTm% Qo(C-T) outlined above).

QoOIP = TotTm% QoC + TotTm% Qo(C-T)

The top five (toughest personnel quality):

The bottom six (easiest personnel quality):

You can see that by handicapping competition quality with Qo(C-T) that Cam Fowler has a tougher quality assignment than both Oliver Ekman-Larsson and Johnny Oduya even though the latter two face tougher competition (higher QoC numbers). In the lower table, you see the opposite effect on Jonas Brodin, whose high teammate quality places him at the same ranking as players with lower quality of competition.

Montreal Canadiens

Viewing the Canadiens defence corps, we see that Josh Gorges and Mike Weaver face the toughest duties from a personnel perspective while Alexei Emelin and Douglas Murray face the easiest.

Defensive zone starting percentage relative to team average (DZSt% rel)

I didn’t find any team stats for defensive zone starts, so I calculated a makeshift version by averaging the DZSt% of all the defencemen on the same team. For a player who has played for more than one team, I included him in the calculation for the team he played the most time for.

DZSt% rel = DZSt% - Average of team’s DZSt%

This isn’t a perfect calculation because it assumes all percentages are equal and doesn’t factor in the differences in games played or ice-time among the defencemen on a team, so it may not be entirely accurate. I performed this calculation to find an individual’s deployment difficulty on his team, attempting to get a metric that would be just as representative of individual deployment as relative Corsi and Fenwick are for individual possession.

Top five (toughest relative zone starts):

Bottom five (easiest relative zone starts):

Brett Bellemore has the second toughest relative defensive zone starts in the league. Perhaps that is the result of a lack of confidence by the coaching staff in the ability of Ryan Murphy, who ranks fourth-lowest in the same category.

Montreal Canadiens

* calculated relative to the defencemen who have played the majority of their minutes for the Florida Panthers

Of the players using Montreal’s average as the benchmark, Gorges is deployed the most heavily in the defensive zone relative to his teammates, while Tinordi has been zonally sheltered the most. PK Subban’s easier zone starts are likely due to the coaching staff wanting to use him in a way that exploits his offensive talents. That is probably also the reason for Murray’s easier zone starts.

Deployment Quality (DQ)

This next step took the rankings for relative defensive zone starting percentage (DZSt% rel) and added in the rankings for the QoOIP stat described earlier. This takes a stat that represents a coach’s deployment of a player and combines it with one that accounts for personnel quality.

DQ = rank of DZSt% + rank of QoOIP

The top-ranked players are those who play against the toughest opponents, often with less than adequate teammates to match the quality of that opposition, and also start a significant amount of the shifts that begin in their own defensive zone.

Institute for the Malevolently Deployed:

The lowest rankings are reserved for players who start an inordinate amount of their shifts outside of their own defensive zone and play either against very-low-quality competition, with very-high-quality teammates, or a combination of the two.

Haven for the Benevolently Deployed:

Montreal Canadiens

This can essentially serve as a window into the mind of the coach in terms of who he believes to be the most defensively responsible on his team. I say ‘can’ because it doesn’t really work in Montreal’s case when two of the top four defencemen in terms of deployment quality are no longer on the roster.

Ranking Players by Possession and Deployment Quality

For this ultimate step in the endeavour to rank the NHL’s defencemen using quality, deployment, and possession stats, we will need some of those possession stats. For that we will use:

Relative Corsi for percentage (CF% rel)

ESGD: CF% rel: Corsi for percentage relative to team's CF% with player not on ice

Range in this evaluation: +10.2 (Mark Giordano) to -8.7% (Justin Falk)
Median: -0.3%

Defencemen Rankings

Now I’m going to combine the rankings for relative Corsi for percentage from among the 237 players in the sample with those obtained above for deployment quality (DQ)

Possession rank = rank of CF% rel + rank of DQ

Once this has been done, we can see who are the

Worst possession defencemen in the NHL for the 2013-’14 season to date (2014-03-27):

These are players who have been given easy competition and/or top quality teammates along with relatively unstrenuous zone starts and still don’t manage to out-possess the members of the opposing team. The Edmonton Oilers shipped two of these players out during the course of the season, perhaps signifying there has finally been a blip of activity on the EEG connected to the Oilers braintrust, although the 225th worst defenceman in the NHL is still a major part of the organization.. There is no justification for such hope for those at the helm of the Montreal Canadiens, unfortunately, as they hold two of the 24 worst possession defencemen in the NHL on their active roster. At least Habs fans can be thankful they do not have three of the worst 24 players in the NHL on their team, as is the case for fans of the Calgary Flames.

Now on to the main reason for creating this (unanticipatedly long) article and look at the

Best possession defencemen in the NHL for the 2013-’14 season to date (2014-03-27):

These are the players who have been facing tough competition in a defensively-minded role all season long but have still managed to achieve good possession numbers.

It’s not all gloomy in the province of Alberta as three of the top five possession defencemen in the NHL play for the Flames and Oilers. The numbers for Mark Giordano are nothing short of spectacular, possessing the best relative Corsi for percentage in the NHL despite having the 14th-toughest deployment quality. EotP's Simon Ledsham thinks the Flames' defenceman has been very good, too, picking Giordano to win the Norris Trophy this season. The New Jersey Devils’ Andy Greene has also been phenomenal, overcoming the 3rd-toughest DQ to have the 27th-best CF% rel.

It’s also interesting that the Buffalo Sabres--the worst possession team in the NHL--have two of the top possession defencemen in Christian Ehrhoff (15) and a resurgent Tyler Myers (22).

Goligoski and Niskanen are very closely ranked among the best, which is fascinating as they were traded for each other in a deal between Dallas and Pittsburgh. That trade also saw a minor league player in James Neal head to Pittsburgh, but I don’t think he ever amounted to much.

I looked into how these players were distributed among the NHL in terms of team CF%. I took the teams ranked in order from best to worst Corsi for percentage and divided the them into five groups of six. I found that the top fifth of the NHL in terms of possession employs eight (8) of these players, the second fifth has six (6), third fifth five (5), fourth (5), and the final fifth or bottom six possession teams have six (6).

After performing the same count with the bottom 30 players, the top fifth, containing the six best possession teams, has six (6) of the worst possession defencemen, the second fifth has six (6), the third fifth three (3), the fourth three (3), and the bottom six teams employ 12 of worst possession defencemen in the NHL.

Montreal Canadiens

Raphael Diaz is the 39th-best possession defencemen in the NHL this season. If he were still on the team, Gorges was healthy, Weaver had still been acquired for practically nothing (probably not without an injury to Gorges), and Beaulieu was rushing up the ice beside Subban where he belongs, the Montreal Canadiens could fill out their defensive corps with players of top-three possession calibre (90th-ranked or better). Instead we regularly see two players on the ice who are outside of even the depth range (7th defenceman, or 181-210th ranking).

Fortunately, the excellent Weaver was acquired after the departure of Diaz (who was also sent away by the Canucks for even less than he was dealt from the Canadiens for; he must have a very serious character flaw or something) and the injury to Gorges. Now to get Murray to stop playing as his defensive partner. Perhaps we can trade him to the Rangers for Diaz. And Anton Stralman (10).

You can compare these rankings to the player usage chart from ExtraSkater I mentioned at the beginning of the article.

Conclusion

I encourage you to look at the spreadsheet to see the rankings of other players you might be interested in. There you may find some surprising positions for other defencemen, like potential Norris Trophy candidate Ryan Suter being ranked number 201 out of 237 or rookie Seth Jones being ranked as the 52nd-best defenceman.

I have created some ‘filter views’ in the spreadsheet where you can easily single out one team’s defencemen for comparison.

I’ve placed the Canadian teams at the top and have also included the best (LA Kings) and worst (Buffalo Sabres) possession teams for comparison of extremes. I think you can actually make your own filter for any team you want to look at, or some other category you desire to compare, by clicking ‘Create new filter view’ in the menu highlighted below.

I plan on revisiting this project at the end of the season to see how the players ended up ranking over the course of the entire season. I know that some players have changed positions (e.g. Chara moving up, Diaz dropping down) in just the week between the first batch of data I collected and a new one collected on March 28th for this article.

Also please feel free to offer feedback (some suggestions on creating a proper DZSt% rel metric would be particularly helpful) and add your own analysis of the obtained rankings in the comments.

I’m also working on these numbers for previous years to see how players fare from year to year. You can see tabs for previous years at the bottom of the spreadsheet, but those are still a work in progress.

I’m also considering making a list of free agent defencemen from these rankings to highlight the best defenders available in the off-season. That shouldn’t take a lot of extra effort.

Now, to perform the same task for the forward position. I hypothesize that the top players in that analysis will be correlated with Selke Trophy nominations, especially centres. I am also interested in seeing where the Canadiens’ forwards rank in the NHL’s possession standings, including Tomas Plekanec and Brendan Gallagher, who I expect to appear near the top with players like Jonathan Toews and Patrice Bergeron. And, like we found for the defencemen, maybe a few surprises as well.