Why do Aggressive Defensemen Experience Sharp Declines at Young Ages?
Alex Pietrangelo’s rough start is an excellent case study of a broader concept.
Alex Pietrangelo’s start to the 2021–22 season has been about as divisive as it gets. Vegas fans and reporters are adamant that he’s carrying the injury-stricken team on his back, admirably taking on a tremendous load and thriving with these heightened responsibilities, while the numbers tell a very clear story of tremendous struggles. When #7 has been on the ice at 5v5, Vegas has gotten massively out-chanced (33.2% expected goal share) and out-scored (30.8% actual goal share). When he’s been on the bench, they haven’t been their usual dominant selves but they haven’t been bad either (51.8% expected goals, 46.2% actual goals). Objectively speaking, the Knights have been much, much worse with him out there than without. Models that adjust for external factors like quality of competition, zone deployment, and teammates are no kinder.
How can we reconcile such an extreme divide? Well, one way would be to ignore the numbers entirely, as some have suggested. But doing that would be a missed opportunity, because Pietrangelo’s start to the season is a good illustration of how skilled defensemen who play aggressively often follow a pattern of negative skew, and why high-end defensemen of his age typically start to decline quickly). This isn’t a hit-piece, and Pietrangelo could very well see his results improve as the season goes along. But the results he’s putting up right now, and how they play into an eye test vs. stats divide, are indicative of something bigger and worth understanding.
Imagine I offer an insurance contract with these stipulations:
- Every day, I get paid $10.
- Every day, there’s a random chance that I have to pay out a premium of $3,000.
- The probability of this premium being triggered is always 0.27% (1/365).
On average, the premium should be triggered exactly once per year, so the expected return of this strategy is $650 per year. I’ll have some bad days where I lose $2,990, and I might even have a bad year where the premium gets triggered, say, three times and I lose $5,350. But overall, I still come out ahead.
But what if a slight tweak is made to the contract that slightly raises the probability of the premium being triggered to 0.55% (or 2 in 365). The expected return of the strategy is now -$2,350 per year. That small change would completely ruin my strategy and lower my expected return by $3,000.
This is a negative skew; the probabilities only need to move slightly in the wrong direction to push your expected returns into the negative. So what does that have to do with hockey?
Many of the best defensemen in the NHL play an aggressive style with a higher risk profile than that of their less talented peers. They gain possession of the puck when their frequent pinches and zone entry challenges succeed, but on the rare occasion that these plays fail, they can be caught high and out-of-position, which leads to dangerous scoring chances against. In other words, they cause good events more often than they cause bad ones, but the bad ones tend to be more severe.
Back to Pietrangelo.
As mentioned above, while Vegas’ injury problems have undeniably taken a toll on the team’s success, Pietrangelo’s struggles relative to his teammates mean we can not simply write off his poor numbers. While Vegas controls shot quantity about the same with or without him on the ice, their control of shot quality is far worse with him. That’s a problem, because control of quality chance shares tends to translate to control of goal shares — and in Pietrangelo’s case, it unfortunately has.
Incorporating data from this season into a robust regression model such as my Wins Above Replacement (WAR), which accounts for competition, teammates, and usage, tanks his overall WAR percentile harder than any other player’s since the start of the season.
While many have said that Pietrangelo is simply being dragged down by his defense partner Nicolas Hague, with whom he has spent over half of his 5-on-5 ice time, the team’s performance when these two have been separated suggests otherwise.
While I understand external factors may be at play here, I refuse to simply ignore the numbers and entertain the argument that Pietrangelo bears zero responsibility for the fact that only 44-year old Zdeno Chara is getting more caved in to start the year.
Fortunately, I think the real story is a lot more simple than it seems. I closely watched Pietrangelo in Tuesday’s 4–0 loss to the Toronto Maple Leafs, and I’ve clipped film that illustrates:
- Why Pietrangelo’s results are so poor.
- Why there’s such a gap between the narrative and the results.
- Why Pietrangelo’s teammates also bear some responsibility for the results.
I believe a good faith film study will help both critics and proponents of Pietrangelo’s game find common ground.
There’s no doubt that Pietrangelo is aggressively involved in the play; nobody out there is trying to do more than he is, and he’s definitely making an effort to carry the team’s offence on his back. That’s what stands out to viewers and fans. But it’s the effect of those efforts that creates the divide between appearances and results.
One thing nobody will disagree on: Pietrangelo’s puck skills are still elite, especially his ability to find pass targets in transition and handle the puck with poise (see 0:09, 0:40, 1:08, and 1:40).
Pietrangelo’s forwards are undeniably failing to cover for and capitalize on many of those aggressive plays (see 0:00, 0:09, and 1:08), but sometimes these aggressive plays succeed nonetheless (0:40 and 1:30). That said, Pietrangelo himself also undoubtedly failed on an alarming number of them, including several in high-leverage situations (see 0:25, 0:30, 0:46, 0:52, 1:01, 1:13, and 1:35).
Overall, there’s plenty of room for common ground. Pietrangelo’s skills are still comfortably NHL-celiber even if the results are not, but it’s clear that the high failure volume on high-risk aggressive plays is less than ideal. If Pietrangelo is making these hyper-aggressive plays solely because he feels that he needs to given the Knights’ injury situation, then it follows that his results will improve as soon as the team gets healthy and he settles down. But if this failure rate is indicative of his having lost a step in terms of skating — perhaps the most reasonable thing to expect for a soon-to-be 32-year old — then something more concerning might be at play.
Now, think back to the concept of negative skew. If the usual age-related decline for Pietrangelo means that he will succeed on just a slightly lesser percentage of the aggressive plays he tries, that alone can make the difference between positive and negative results.
Aggressive Defensemen in General
If you look across the league at other talented defensemen who play a high-leverage, aggressive style, you start to see a pattern emerge. The chart below displays the RAPM-isolated impact that Pietrangelo, Drew Doughty, Erik Karlsson, and Seth Jones have had on a couple of on-ice stats:
Expected Goals For and Against (xGF and xGA) are measures of shot quality, while Fenwick For and Against (FF and FA) are measures of quantity. Across the board, their offensive and defensive impact on shot quantity are both higher than their impact on shot quality. As these players decline, the positive events may still occur more frequently than the negative ones (which will be reflected in shot quantity and perhaps the eye test), but the return from these events will turn negative (which will be reflected in shot quality and should ultimately be reflected in goals) because the bad events are more severe than the good ones, and the frequency no longer outweighs the severity.
Not only have all three of the players listed alongside Pietrangelo seen their underlying results decline sharply at a younger age than expected, but all of them have put up at least one season with some of the worst underlying metrics in the league. If he does not turn his game around quickly, the 2021–2022 season will be Pietrangelo’s turn to do so.
I believe the reason these defensemen have declined so severely at such a relatively young age can be explained by the concept of negative skew. Their positive impact is dependent on succeeding on a high enough percentage of their aggressive plays that they offset the big losses they suffer when their risks fail. In these high-leverage situations, it does not take a large drop in success frequency to see a large drop in expected returns. For an aggressive defenseman, losing just half a step in their stride, or half a second in their reaction time, can make all the difference.