It’s been a turbulent season for William Nylander. A protracted contract negotiation saw him miss the first two months of the season, with fans and media members taking turns to criticize his character and value to the Leafs. When he returned, he entered the worst slump of his young career from a scoring perspective, and was predictably rusty, which only compounded the criticisms.
However, as we would expect, Nylander didn’t suddenly forget how to play hockey over the offseason. He got back up to speed and has been a noticeably improved player in many aspects when compared to last year. Since December 6th (the day of his first game back), his 3.35% CF Rel% ranks second among Leafs forwards (Andreas Johnsson is first). If you look at more complicated stats, he looks even better. Nylander’s impact on shot generation stacks up with some of the very best players in the league.
One stat I like to use to assess this is called Regularized Adjusted Plus Minus (RAPM), which can be found at evolving-hockey.com. RAPM is an all-in-one number that measures how a player drives play. The details of how to calculate RAPM are in the link above, but the short version is that it is a measure that estimates a player’s impact on a play driving metric (such as goal generation, shot generation, shot suppression, expected goal generation, or expected goal suppression) accounting for their teammates, their competition, their score usage, and their zone usage. It is not a perfect encapsulation of a player’s play driving ability, but in my opinion, it is one of the best publicly available stats out there at accounting for usage and assessing which players tilt the ice in their team’s favour. Below, I’ve plotted the densities (essentially, the frequency) of players’ impacts (on a rate basis) on shot generation for (CF60), expected goal generation for (xGF60), shot differential (CD60) and expected goal differential (xGF60), as calculated by RAPM, highlighting where Nylander stands relative to the league. The x-axes in this case represent the impact provided in each stat relative to an average player, per 60 minutes.
We should keep in mind that these estimates of player impact are just that — estimates. There are invisible error bars around each of those blue vertical lines, which can be significant. Nonetheless, I think that we’d all agree that these results make sense to us. Nylander is an outstanding passer and transition player, so it makes sense that his impacts on shot generation are well above-average, which drives his excellent shot and expected goal differentials on the whole. He grades out as average defensively by these measures (not pictured) which also makes intuitive sense. These play-driving numbers have Nylander knocking on the door of the truly elite players in the league. Which is to say, from the perspective of tilting the ice in his team’s favour, it’s hard to ask any more of Nylander.
However, the elephant in the room is that play driving, while a critical part of hockey, is not the target. The target is scoring goals, and unfortunately for Nylander, the Leafs haven’t done a lot of that when he’s been on the ice. The Leafs score 2.75 GF/60 at 5v5 with Nylander on the ice, putting him seventh among Leafs forwards. As you would expect given his elite play driving numbers, this is primarily the result of a low personal and on-ice shooting percentage. The Leafs as a whole shoot just 8% at 5v5 with Nylander on the ice (per Natural Stat Trick), which is not terribly far from the league average, but is notably below the team average. Furthermore, Nylander himself has only shot 4.8% at 5v5 this year, beating only Connor Brown and the since-departed Par Lindholm among Leafs forwards. Per MoneyPuck, Nylander’s 5v5 goals per minute is only 44% of his expected goals per minute. This means that accounting for shot quality, a shot taken by Nylander this year is less than half as likely to go in than if it was taken by an average forward.
The common refrain is that he is likely to see his shooting percentage bounce upwards, and that he’s too talented to shoot 5% forever. That is almost certainly true. One of the most important and fundamental discoveries in the early days of hockey analytics was understanding just how volatile shooting percentage can be, and that the vast majority of players fit into a narrow band of shooting performance. Outlying results like this are not likely to persist long-term.
However, the troubling aspect is that this is the third straight year that Nylander has underperformed his expected shooting percentage. At some point, we have to stop expecting his shooting to revert, and consider the possibility that he’s simply not an effective shooter. While his shot seems good superficially, he has at times struggled with getting it on net. Over the course of his career, just 52% of Nylander’s shot attempts have ended up on target, with the league median for forwards in that time being 58%. Even when his shots make it on target, only 7.5% of them go in, compared to the league median of 9.25%. At a certain point, the eye test takes a back seat to three years of shooting data.
From the Leafs perspective, this is problematic in that his upside is capped. It’s hard to become an elite point scorer while being a below average shooter. That Nylander is still 63rd league-wide in 5v5 P/60 since his rookie year despite this is a testament to his elite passing and the fact that he has shared the ice with the best 5v5 goal-scorer on the planet. Furthermore, Nylander won’t drive goals to the same degree that other elite offensive players do, despite his ability to drive shots and expected goals, simply because he doesn’t convert his chances at a high enough rate.
The Leafs have a partial remedy to this, which is to keep him tethered to Auston Matthews. Matthews is one of the best shooters in the league, covering for Nylander’s weakness in converting chances, and Nylander, as one of the best play drivers in the league, helps Matthews spend more time in the offensive zone, where he’s at his best. Matthews is one of those special players who will always be better than their play driving metrics say they are. Every game, we have more and more evidence that Nylander is the opposite.
This is a real tragedy, because Nylander’s offensive game outside of his shooting is elite. He has the potential to be one of the very best offensive players in the world, given his transition ability and passing. Manual tracking data suggests that Nylander is among the league leaders in creating shots for both himself and teammates, as well as creating zone entries. If he was an average shooter, his 5v5 scoring over the past three years would be similar to Nicklas Backstrom’s — another Swedish pass-first player who is the second banana to an elite goal scorer. Backstrom will probably end up in the Hall of Fame. The main thing separating he and Nylander’s 5v5 scoring over the last few years is shooting percentage.
If you’re looking for some indication that Nylander’s shooting is salvageable, the man to look at is Nathan MacKinnon. Through his first four years, MacKinnon’s shooting percentage was 8.1%, and there was discussion over why such a can’t-miss prospect was not producing the way we would expect. Over the last two, his shooting percentage is 12.4%, and suddenly, MacKinnon is a consensus top-10 forward in the world. It’s also not as if MacKinnon’s expected shooting percentage drastically shot upwards either. It’s higher in the last two years than in the previous three, but only by 0.5 percentage points. Furthermore, his expected shooting percentage is lower than Nylander’s has been across his entire career (one of the most annoying myths about Nylander is that he’s a perimeter player who takes low percentage shots). Some time in the last few years, whether by a change in his game or random variance, MacKinnon figured out how to shoot, and changed from a frustrating young player with potential to an elite offensive player. It’s not out of the realm of possibility that Nylander can do the same.
If I ran the Leafs, I’d be doing everything in my power to figure out why Nylander’s shooting has produced such poor results, and what (if anything) can be done to fix it. He’s a brilliant player already, but if this (potentially) low hanging fruit is plucked, he becomes a world-beater. More generally, a team that is able to coax out a percentage point or two out of a player’s shooting percentage would be unlocking a massive competitive advantage. Players who can sustainably convert a high proportion of their shots to goals are unbelievably valuable in a league as tight as the NHL. If I were running an analytics department, it’d certainly be one of my areas of intense investment. The Leafs are fortunate enough to have two elite shooters already in Matthews and John Tavares. Their goal this offseason should be to try and see if there’s a way to internally develop another.
RAPM data is from evolving-hockey.com, and includes all even strength data
All other information is from Natural Stat Trick (naturalstattrick.com) or MoneyPuck (moneypuck.com)
Stats are current up to March 22