Size Matters: Should Getting Bigger Be a Priority For Teams?
We talk about size a lot in hockey. We debate whether or not guys like John Scott (6-foot-8, 270 pounds) can “patrol the ice” and make an impact as a goon. We’re debating the merits of Stu Bickel. We say young players need to bulk up. We say things about teams as a whole like, Los Angeles plays heavy hockey or The Wild need to get bigger.
It’s this last one that got me started. The Wild need to get bigger. If you pay attention to Minnesota Wild hockey, you’ve heard that a lot over the last year or so, particularly about the team’s defense. Hell, you’ve heard it this week from people who believe Bickel belongs in the Wild lineup or when his move to wing on Thursday is discussed.
This bit of folk wisdom — essentially, bigger is better — has been bandied about frequently by Wild fans, yet the Wild were solid defensively with a 2.42 goals against average (GAA), which ranked 8th in the NHL. They did that without a consistent goaltender and with the lightest defenseman in the league — Jared Spurgeon — seemingly proving that size and grit come in a distant second to actual skill.
Also, the Wild got smaller when they lost 6-foot-4, 216 pound Clayton Stoner, but I’d argue they got better with his departure. If that’s true, they got better by getting smaller.
I started to wonder if this bit of folk wisdom is actually true and if it is true, if it is of any value to a general manager who is constructing a team.
And here we are: Does size actually matter? And if size matters, how does it matter?
I’ll explain the method as we go, but I’m looking for a couple things.
In general: Does a larger team defense prevent more goals or have better possession than smaller team defenses. And, if so, as individuals, do bigger defenseman prevent more goals than smaller defenders and do they do better in terms of possession.
For the Wild: Were Wild defenders actually smaller than their counterparts on other teams? How does their size stack up and is there any evidence that says they would have been better last year if they were bigger?
What Is Big?
I’m using weight as a measurement of size. There’s a correlation, unsurprisingly, between height and weight among NHL defenseman (and probably other humans too), so there’s reason to equate them and I believe that weight is a better indicator of what we’re looking at — what size adds to a defense — than height. But since I have the stats and it’s worth being sure, take a look below.
This is every defenseman in the NHL from the 2013-14 season who played at least 50 minutes, from Spurgeon as the lightest defender on the left, up through Zdeno Chara in the upper right. This should all probably go without saying, but I wanted to be sure that it was clear and true that weight can be a stand-in for the sometimes abstract “size.”
Adding It Up
Let’s look at overall team size.
A little explanation of the numbers. For ease I’m going to call the weight number I’m constructing WAD, the weight of the average defender. This is calculated by taking the weight of every defender who played at least 50 minutes last year and averaging their weight by team while taking into account how many minutes each defender played so that we come closest to the weight of a defender on a given team in any given minute during the 2013-14 regular season.
The other stat here we’ll get to in this short series of graphs is Corsi For % (CF%), which here is the CF% for all defenders on a team who played at least 50 minutes.
Here’s the breakdown by team, ranked heaviest to lightest. The Wild were in fact smaller than most. They were the second lightest team in the league.
[table id=10 /]
In this first chart I’m plotting WAD vs. the team’s GAA. The yellow dots are playoff teams and the red dots are everyone else.
The trend line shows a fairly strong correlation between the team’s size and a lower GAA. Though, there are a couple of outliers that are notable here. In the lower left you’re seeing the Wild, who have the second lowest WAD with an average weight of 195.964 pounds. Only Calgary had a smaller defense at 194.062 pounds. The Wild’s GAA was better than most teams in their size range. The other notable one is the dot above 3 goals against furthest to the right, that’s Ottawa. They’re the sixth biggest defense at 211.195 pounds with a pretty miserable GAA of 3.15.
Unsurprisingly most of the playoff teams are found in a lower GAA range, but the yellow dots are more heavily weighted toward the right side of the graph, with eight of the league’s ten heaviest teams making the playoffs and having pretty solid GAA. On the other end, only two of the league’s ten lightest teams made the playoffs — the Wild and the New York Rangers.
Next, let’s take a look at WAD vs. the total CF% for each team’s defense. Again, playoff teams are in yellow.
Another moderate correlation, though it is less pronounced than the comparison with GAA.
This is a different conversation and has little to do with weight, but a little fuel on the fire is never bad: The blue dot with the best CF% here is the same team as the red dot with the lowest GAA in the previous graph. The New Jersey Devils. Not a huge team, but solid GAA, solid CF% in their defensive corps. A notable exception in these charts. Hi, how are you doing shootouts?
Right now we’ve seen that the bigger teams tend to have lower GAA, a slightly higher CF%, and are making the playoffs at a higher rate than smaller teams.
So, a team who wants to be better should look for bigger players, right?
The Size Have It
Signing a player isn’t done as a defensive unit but on an individual level. Let’s apply the same test to individual players that was placed on teams above.
I’m taking each of the players that’s included in the statistics above — every defenseman who played at least 50 minutes last season — and mapping two things. I will map their weight and Goals For% (GF%), then their weight and their CF%.
First, Weight and GF% for individual skaters.
The trend line shows a very slight inverse relationship between a player’s weight and their GF%. It’s slight enough that, while it exists, I’d say that it’s not really all that significant. There are a few outliers that are so far out I feel we should mention them: Brandon Gormley (73 minutes played) wound up with a GF% of 100% and at the other end Philip Samuelsson (78 minutes played) and Xavier Ouellet (58 minutes played) had a GF% of 0%.
I graphed the same results for all players who played a minimum of 200 minutes and the trend line didn’t change much. So it’s not entirely thrown off by players like Ouellet and Gormley. There’s some consistency to this trend line, for whatever that’s worth. I still believe that the trend line is slight enough that there isn’t much for a GM to work off of in terms of these statistics being useful for constructing a team. The most useful thing to note is simply that it doesn’t parrot the results from above of larger team defenses being better. On an individual level, so far, those results don’t hold true.
Now, Weight and CF% for individual skaters.
A similar result. The trend line shows an inverse relationship, with fewer outliers, but the results aren’t something a GM is going to build his team around. However, it certainly shows that the inverse isn’t true. (The inverse being that bigger players are better than smaller players)
The Touches in California
Stephen Burtch had an interesting post at Sportsnet a few weeks ago where he proposed a new way to look at a defenseman’s defensive capabilities. As a starting point Burtch uses Corey Sznajder’s summer research on zone entries and a statistic he calls Touches. Burtch defines it this way: “[Sznajder] defines a Touch as a situation where the defending player gains control of the puck with the intention of exiting his defensive zone.”
Burtch expands upon Touches saying as a lead-in to his own statistic:
Shot-attempts against (also referred to as Corsi Against) are a negative outcome for any defensive unit, because any attempt—shot quality aside—could theoretically end up in the back of the net. The two key shot-attempt points to consider are: 1) the defensive unit does not have control of the puck; and 2) the offensive unit does and is in a position to potentially score a goal, no matter the actual likelihood.Considering these two points, it’s logical that defensive-zone touches and shot-attempts against are at odds with one another. If we can describe the ratio between the two events, we can track a blueliner’s defensive skill when the puck is in his zone.
Burtch creates a statistic he calls Touches/CA. That’s Touches over Corsi Against, where a good defender has more touches to clear the zone than Corsi Against (1.00 and above) and a poor defender has a lower Touches/CA (below 1.00 would be quite poor).
It’s an interesting proposition of a way to judge pure defensive skills and finding a way to cut out offensive positives as a positive on the stat judging defensive skill. Let’s use Touches/CA against weight and see if there’s a different outcome than the slightly negative correlation we saw between size and GF%/CF% above. (NOTE: Burtch’s data only includes defensemen who had at least 450 Touches. So the sample size from the previous examples is reduced here.)
We’re seeing basically the same negative correlation, where the bigger the defender the more likely they are to have a bad Touches/CA. But, like with the other graphs, the trend is very slight and while there’s a statistical interest that they all flow backward of what our folk wisdom tells us, the trend is slight enough that no general manager should base a decision on this.
The Folk of Folk Wisdom
An assessment of size in individual players and in teams has slight, but opposite conclusions. A bigger defenseman isn’t likely to be better than a smaller defenseman at large, but a bigger team is likely to be better statically than a smaller team, at least in terms of GAA and possession.
If you try to interpret these stats in a vacuum in order to take action, it says very little. Your team should be bigger, but as you find the individuals to put on that team, they shouldn’t necessarily be bigger. Or you need bigger players, but getting bigger players isn’t necessarily the best way to do that. Right? Wait. What?
As I’ve repeated, it seems obvious that a GM shouldn’t base a decision on the size of player. There’s very little reason to use size as the sole determining factor for a signing, but we do see that. While he’s not a defenseman, can you argue that John Scott would still be in the NHL if he was 5-foot-10? Size is used as a determining factor at times and there isn’t any evidence that a bigger player is better where it counts or has the ability to make the team better.
The armchair GM isn’t wrong when s/he says a team needs to get bigger, there’s reason to believe that could help the team. But the reality for an actual GM whose fingers aren’t covered in Cheeto dust during a game is far more complicated than that. It’s great if a team can get bigger, but size will never be a replacement for skills.
NOTE: This is only looking at stats from last season because of the questions that were initially asked and my focus on the Wild last season. There will be a follow up article with an expanded sample size to see if these trends persist over time.
All stats pulled from HockeyAnalysis.com and HockeyReference.com, except for all Touches/CA data.