I’m taking a small detour from my original planned posting schedule to do an addendum to a previous post. Previously I looked at how NHL teams deploy their players differently based on size and the strength state of the game and took a look at how the league has trended since the 2007-08 season. At the time, I noted differences between the power play and penalty kill in height and weight and added that I suspected this was due mainly to the increased percentage of ice time taken up by defensemen. Since my planned method for analyzing playoff success is more time consuming and I was waiting for the Final to end, I decided to look into my assumption and break ice time distributions down by position as well. This will be a relatively short post.
Forwards Versus Defense
I kept my methodology simpler for this exercise than in part two. Instead of comparing size and age at each strength state with the all situations weighted mean, I just took the weighted mean for each state directly. I also included confidence intervals for each weighted mean (weighted by ice time) so that I could see whether the differences between strength states and positions were significant. Because of the use of confidence intervals, I excluded draft position because the large number of undrafted players compared to late round players give the data a bimodal distribution. To be honest, I’m not completely sure (confident) that this is the correct way to use confidence intervals, but I don’t think it’s wildly wrong either.
Each dot in the plot above is the mean for that population and the lines extending from it represent the 99% confidence interval for the mean. The means can be said to be different at with 99% confidence if the confidence intervals do no overlap. Looking at age, for example, both forwards and defense are older on the PK than on the PP or at even strength. However, there is not a significant difference in the ages of forwards and defense on the power play.
Looking at the general difference between forwards and defensemen, it’s pretty clear that defensemen are taller, heavier, and denser at all strength states, and even older on average, except on the power play. Defensemen also have a much clearer stratification between strength state with penalty killing defensemen being older, taller, heavier, and denser than either other strength state. Defensemen deployed on the power play are also shorter and lighter than defensemen at even strength, though not significantly younger or less dense.
Forwards, by contrast, are much less stratified. There’s no significant difference between forward height or weight at any strength state. Penalty killing forwards are older on average than even strength or power play forwards, but also have a lower BMI, which is intriguing. I wonder if this is explainable by the need for forward penalty killers to be faster, and therefore less dense, if not less heavy.
In all, this seems to support my conclusion that penalty kills were heavier and taller mainly due to the presence of defensemen. It’s harder to reconcile the difference in age on the power play, since the previous post suggested that the power plays skewed younger than even strength, while this suggests that forwards are older on average. Looking at my previous post, I’m not sure why I determined that power plays were younger, since it looks like it’s actually similar to even strength, just with wider variance. The chart here backs that up, with forwards being slightly older and defensemen being slightly younger and neither escaping the confidence intervals.
Season-To-Season Trends
I repeated the process above for each season, once again calculating 99% confidence intervals for each weighted mean. Since the number of players examined in each season was significantly less for all seasons, this meant the confidence intervals were much wider than before. In fact, it is difficult, but not impossible, to find differences between the means at the season level.
There’s a lot going on in the chart, so it’s easiest to just go through it from top to bottom. Forward have trended younger, but only the difference at even strength is significantly different at the 99% confidence level from 2007-08 to 2020-21. As noted in part 1 of this series, the trend seems to have reversed in recent years. Penalty killing forward age is also one of the few clear instances where the difference between strength states is significant at the season level. Defense follows the same pattern of getting younger, but not significantly so, with the trend reversing recently.
Forward height has been remarkably consistent since 2007-08. Defenseman height has also been quite consistent overall and has actually trended up a little on the penalty kill, but not significantly so. Defensemen are also significantly taller than forwards at even strength and on the penalty kill across all seasons.
Forward weight is the first metric where there is a significant difference at all three strength states from the beginning of the period to the end. It’s also interesting to see that the differences between the strength states, although never significant to begin with, seem to be shrinking and in 2020-21 they were almost identical. Defense has also gotten significantly lighter, though the difference between the power play and penalty kill has remained consistent at the season level. Defensemen on the power play in particular have gotten much lighter in the last couple of seasons.
Lastly, forward BMI has followed, essentially the same trend as forward weight, which is an intuitive result given that height has been pretty stable. The trend is roughly the same in defensemen, but steeper. In fact, decreasing BMI in defensemen has been the strongest trend in the league over this time period. This is likely a reflection of the end of immobile stay-at-home defensemen and the rise of smaller, faster, puck moving defensemen.
Conclusions
Although this post wasn’t originally planned for now, I am glad I took the time to put it together. It helped confirm some of my suspicions and re-evaluate another one. It’s also helpful to understanding to see a breakdown of some of the aggregated numbers from earlier and will have some influence on the course my future posts take. I’m not sure whether I will finish my playoff analysis or individual contributions first, but either way, it should be up soon.
Fin.
A huge thank you goes to Evolving-Hockey for providing team, player, and even play-by-play data in a clean, easy-to-use format. When I did my last major project I was combing through the NHL play-by-play data up till 2017-18 and it was quite messy. Fortunately, Evolving-Hockey absolutely has the best quality of any of the major hockey stats sites I’ve looked at and it makes hobbyist work like this so much easier. For that reason, although I’ll use their data and make some of my data available, I won’t include anything from their site that can’t also be found on NHL.com or other free sites.