Holding the line: Are big defencemen better at controlling the blue line?

    Andrew Anstey

    One of the more divisive moves in the first round of the 2016 NHL entry draft was Winnipeg’s selection of Logan Stanley at number 18. It seems that Stanley was specifically targeted by the Jets, who traded pick 22 and 36 to acquire pick 18 and 79 from Philadelphia. While Stanley was seen as a consensus mid to late first round pick prior to the draft, there are concerns that he was drafted solely for his size and is lacking in offensive talent. Stanley is an absolutely towering figure at 6’7″ and 225 pounds, but put up only 5 goals and 12 assists in 64 games in his draft year.

    While Stanley could very well turn out to be an excellent NHL player, this move is reflective of an old school NHL desire for size. Big, tall players have long been targeted and lauded by NHL scouts and GMs – this is especially evident in defencemen. Large defencemen are generally assumed to have superior upside in the defensive aspects of the game, due to their long reach, strength and ability to throw big hits. According to common sense, this is a pretty reasonable viewpoint – big players should have a longer reach to break up passes and zone entries, and can use their weight and reach to control scrums along the boards.

    Zdeno Chara fits this bill perfectly – the 6’9″, 250-pound behemoth emerged in Boston as one of the premier defensive players in the league. Old school and new school hockey minds both agree on Chara – in his prime, he absolutely dominated at puck possession. Watching him play, he seemed impossible to get past one-on-one due to his incredible reach, and delivered crushing hits on a daily basis. It was clear that he used his size to his advantage to limit scoring chances, break up rushes and keep control of the puck. Chara’s excellent puck-possession numbers make sense: it is well established within the hockey analytics community that limiting controlled zone entries is a key way to limit shots on goal. On average, controlled entries tend to generate more than double the amount of shots that dump-ins generate.

    But does this hold true universally, or is Chara a special case? Are big defencemen better than smaller defencemen at controlling the puck and preventing zone entries? Does it make sense to target big defencemen at the draft on the assumption that their size will give them a higher ceiling defensively?

    I used zone entry defense data from Corey Sznajder‘s All Three Zones project to see if there was any consistent relationship between a player’s size and their ability to limit zone entries. In an amazing effort, Corey tracked every single game in the 2013-2014 season and recorded every single zone entry. We can use this data to see not only which forwards were good at bringing the puck in to the zone with control, but also which defencemen were able to break up plays at the blue line and force dump-ins. I matched Corey’s data set with height and weight stats from NHL.com‘s real time stats, as well as possession stats from corsica.hockey. I limited the study to defencemen with at least 200 minutes of even-strength playing time who were targeted on at least 100 zone entries. This gave us a sample of 197 players to work with.

    First, let’s take a look at the size of the players in the sample. The height ranged from 68 to 81 inches, with most players generally falling between 72 to 76 inches (6’0″ to 6’4″). From the histogram below, we can infer that the data is roughly normally distributed. The data set had a mean height of 73.73 inches (~6’2″), with a standard deviation of 2.05.



    Based on the mean and standard deviation, I established 3 groups of players, which I will refer to as “small”, “medium” and “large”:

    Small: Players more than one standard deviation below average height (less than 6’0”)

    Medium: Players within one standard deviation of average height (6’0” to 6’4”)

    Large: Players more than one standard deviation over average height (over 6’4”)

    I looked at how each of these three groups performed in four categories:

    Carry %: What percent of times did an attacking player enter the zone with control against this defenceman?

    Break-up %: What percent of times did the defenceman successfully break up a zone entry?

    Dump %: What percent of times was the attacking player forced to dump the puck when targeting this defenceman?

    Prevent %: Sum of break-up and dump-in percentage (ie. the attacking player did not enter the zone with control of the puck). The reverse of carry %.

    Here’s how each group stacked up:


    Statistically, there is no significant difference between any of the groups in any of the categories. Small, medium and large defencemen all performed equally in terms of their ability to control entries into their defensive zone. Thanks to Corey’s work, we also know the expected value for each type of event in terms of shot rates. From Corey’s data, Dominik Luszczyszyn of The Hockey News calculated that on average, each controlled zone entry (carry) generates an average of 0.66 shots, while a dump-in generates only 0.29 shots on average. This agrees with Eric Tulsky‘s (now of the Carolina Hurricanes) initial work on zone entry research, which involved a similar study of a much smaller set of games. If we make the assumption that a break-up results in 0 shots against, we can establish a value for each player based on their zone defense ability:

    Expected shots per entry = 0.66 * (Carry %) + 0.29 * (Dump %) + 0 * (Break-up %) 

    Once again, all size groups performed almost identically based on this metric – small and medium players averaged 0.49 shots per entry, while large players averaged 0.48 shots per entry. Next, I wanted to see how much correlation existed between the different variables in my data set. Do heavy players give up more zone entries? Do players who prevent zone entries tend to reduce their overall Corsi against? The table below sums up the correlation between each of the factors.


    I determined the R-squared value between each variable. R-squared tells us how much of the variance in one variable can be explained by variance in the other variable, or in other words if there is any dependence between these variables. It doesn’t tell us whether there’s a positive or negative relationship, just that one exists. Significant relationships are highlighted with bold text and a graded green background depending on the R-squared. A few variables emerge as being significantly related to each other based on this analysis. First, we can see that there is a strong relationhip between player height and player weight (duh). Otherwise, height and weight have no relationship to any other variables, including Corsi for and against. In other words, we can confidently say that being big and tall has no effect on how defenders perform in terms of limiting zone entries, puck possession, shots against, or their own point production.

     We can also ignore the relationships between carry %, break-up %, dump % and success % due to the collinearity of these variables; we know these variables are dependent because of they way we calculated them. We can see that points per game played was strongly related to relative Corsi for per 60 and relative Corsi percentage – this is not surprising, as it is both intuitive and well established that players who generated more shots and puck possession will generally score more points.

    There is an apparent correlation between a player’s zone defence (carry %, break-up %, dump % and success %) and their ability to limit chances against – we can see that there is a relationship between each of these variables and relative Corsi against per 60. This relationship affects relative Corsi % by extension. The figure below shows that there is a relationship between success % and Corsi against per 60. We can see that players with a better success % tended to have less shot attempts taken while they were on the ice. This is in agreement with previous zone entry work: controlled zone entries produce more shots than any dump-ins and failed entries, and thus players who can limit successful zone entries should be able to reduce shot attempts while they are on the ice.


    Interestingly, there is also a weak positive relationship between points per game played and break-up %. Why do we see these relationships? It’s possibly because both of these measures indicate a certain level of player skill in terms of puck handling, stick work, positioning and “hockey IQ”. It stands to reason that a defender who scores a lot of points will tend to be a smart player with puck handling and passing skills. You can also assume that a defender with above average “hockey IQ” and puck-handling skills would likely also be above average at breaking up zone entries with good positioning and stickwork.


    Finally, we can see a very weak relationship between relative Corsi % and targets per 60. These two variables may not seem related, but on closer examination it makes a bit of sense. Looking at the figure below, we can see that this is a negative relationship – players who were targeted more often generally had worse puck possession while they were on the ice. In order for a defender to be targeted on a zone entry, the other team needs to have the puck. Thus, if a player is being targeted very often, it stands to reason that the other team must also have the puck more often when this player is on the ice. While this relationship is very weak, there may be a bit of substance to it.



    To summarize, by comparing zone entry data to player size and possession metrics, we were able to reasonably conclude a few things:

    • Tall players don’t seem to be significantly better than smaller players at preventing zone entries or forcing dump-ins, despite conventional wisdom that big and tall players tend to have higher defensive upside due to their size.
    • We can see the effects of zone entry defence on an individual level; players who demonstrate talent at limiting successful zone entries also tended to limit shots attempts against. This agrees with previous zone entry research and team-level correlations.
    • Defencemen who score more points also tend to be slightly better at breaking up zone entry attempts. This may point to the fact that players with good scoring ability have good stick skills and/or positioning that also translates to defensive talent at breaking up plays.

    Players don’t necessarily excel defensively because of their size and reach. When evaluating defencemen, it is crucial to value their skill level and observed ability to break up plays rather than size alone. Players like Logan Stanley may turn out to be excellent NHL defencemen, but size is only a small attribute in their game – you don’t need to be a monster to have true defensive skill.

    • Great stuff, Andrew.

      It’s an interesting topic of real, practical value, and it’s fascinating how you built on existing data and knowledge, like Corey’s zone entry work.

      Also, that’s a great use of charts to illustrate your points, and break up the reading.

      Speaking of breaking up the reading, you tackled a lot here, and could have broken up the analysis into three separate, digestible topics that we could have discussed individually.

      Question: can you think of some specific examples that illustrate each of your points? We all know Chara pretty well, but it would be interesting to take a closer look at some other specific tall defensemen.