Picking the 2019-20 NHL Awards Using Analytics - Part 3: Norris Trophy (Patreon)

Is there any NHL award debate as vicious as the Norris? Because stats that actually measure defensive play are mostly unknown to anyone but a small subset of hockey fans, Norris arguments are almost always a swirling hellhole of points and hearsay that usually end in hurt feelings and nobody changing their minds. Comparing defencemen is so difficult that even some of the advanced metrics we commonly use can be misleading as well. Here I'll try to address some of those issues, taking a bit of a holistic statistical approach to the question of "who was the best defenceman of the 2019-20 seasons." That doesn't mean "points and plus-minus matter!", but it does mean considering a broad set of analytics to resolve this very tight contest.

Note: I am setting parameters of reasonable-ness on these picks though: the winner of the Norris must play top-pairing minutes. Call me old-fashioned, but the guy voted best at his position in the league should probably play a prominent role on his own team. (Sorry Adam Fox).

Other Note: The stats used in this piece are based on EvolvingWild's GAR, xGAR, and RAPM models which are available to their patreon subscribers at evolving-hockey.com. Stats on player competition can be found at PuckIQ.com/woodmoney.

Methodology (Skip this if you don't care and just want to see my picks) 

When I make player cards for defencemen, the main stat I use is EvolvingWild's xGAR and xWAR as opposed to GAR and WAR. Why the "x"? GAR is based heavily on goals for and expected goals against. This is a great way to evaluate offensive and defensive play because it does a good job of including the finishing talent of a player without penalizing them or rewarding them for the performance of their goalie. But when we use it for defencemen, there are some issues - Victor Hedman has less to do with shots going in than Steven Stamkos does. So xGAR, which subs out goals for with expected goals for is better. 

But there's another issue that arises: xGAR, to improve its expectation ability, includes an estimate of shooting talent. Once again, that's all well and good for forwards, who take enough shots and score (or don't score) enough goals that we can reasonably project it. But for defencemen, there is almost no year-to-year correlation between shooting talent - between 18-19 and 19-20 shooting xGAR it was less than 1%. For example, Morgan Rielly went from 12 shooting xGAA in 18-19 (highest by far among D) to -2.5 in 19-20 (292nd). Even more damning, if you just used xWAR to name a Norris winner last year, you would pick Rielly who finished first in the league in that category. But a whopping 78% of his xWAR was based on "shooting talent" alone, which we know from this year was mostly random luck. 

So I'm going to avoid the temptation to just use WAR or xWAR, sort from first to last, and call it a day - as easy as that would be. Instead I'm going to use a combination of RAPM stats (play-driving stats adjusted for context) and xGAR - Shooting.

Who's Not on My Ballot:

John Carlson - RD, Washington Capitals

Carlson is clearly a gifted offensive defenceman, as evidenced both by his great offensive stats and poor defensive ones. His hockey-card numbers - especially point totals - are impressive, and will get him a lot of love from mainstream voters. But his underlying numbers on offence aren't elite, including his even strength xGAR once adjusted for shooting. A career year for Carlson, and one he'll be remembered for, but not one of my Norris picks.

Honorable Mentions

Dougie Hamilton - RD, Carolina Hurricanes

Sigh. Dougie would be the winner in a walk if it weren't for an injury that limited him to 47 GP. I thought about it, but I just can't justify breaking all precedent on his behalf. Hamilton is an elite driver of shot attempts, goals, and expected goals on offence, but his defensive game is rock-solid as well - especially on the penalty kill. He may not get the respect he deserves in mainstream circles because of an unearned reputation for one-way play, but Dougie was the toughest cut on the list.

Ryan Ellis - RD, Nashville Predators

Ellis, like Hamilton, has his case killed by an injury. Before that, he was putting together a very strong case (albeit one bolstered by some puck luck). He's excellent at generating and suppressing quality chances, good on the powerplay, and excellent on the penalty kill. Hopefully he can come back next season just as good.

Alex Pietrangelo - RD, St. Louis Blues

Pietrangelo had a hell of a contract year. He's a superb offensive player across the board and capable defensively (albeit more at suppressing attempts than quality). He plays a huge role on a Cup contending team and faces very difficult assignments. It was extremely close between him and the third place player on my pretend ballot, and Blues fans can be very proud of the season their captain had - even if they might be a little nervous about what it means for their cap situation.

My Nominees

#3: Roman Josi - LD, Nashville Predators

Only two defencemen in the league were relied upon to the extent that Josi was, but that huge role did not have any negative effect on his play. Josi is dominant offensively, a shot attempt and transition machine, and a strong driver of goals for. His defensive game is also fairly strong, and the fact that he prevents quality more than quantity suggesting that he forces opponents to the perimeter. His two-way game is not quite strong enough to elevate him to the top two, but this is an incredibly impressive season for the Predators' captain. 

#2: Jared Spurgeon - RD, Minnesota Wild

Coming off a big (but not big enough) new extension, Spurgeon once against delivered a spectacular season. At even strength, he's elite at both ends of the ice, driving shot attempts for and preventing quality shots against at exceptional rates. He's a strong powerplay asset, despite being weak on the PK.

So why not Spurgeon at #1? A few reasons. Spurgeon is statistically so close to my #1 choice that we start to get picky at the margins. He played the 4th most minutes for the Wild this year, missed seven games due to injury, and ranked only 36th in penalty impact. Finally, his high shot rate and shot attempt rate and lower xGF suggest a tendency to force pucks to the net from the point on low-percentage shots compared to the winner. It's a neck-and-neck race, but ultimately he's the runner-up in my books. 

#1: Charlie McAvoy - RD, Boston Bruins

If it weren't for a lack of gaudy point totals, McAvoy would be recognized for the superstar that he is. He's a two-way defenceman in the purest sense: almost no other blueliner at his level defensively comes even close to his offensive ability. Despite the lack of box-score production, McAvoy drives goals, expected goals, and shot attempts at a high level. He led all defencemen in shooting-adjusted xGAA this season.

Finally, McAvoy drew penalties at by far the highest rate among defenceman in the league. He does all this as the Bruins' #1 defenceman playing the 14th toughest minutes in the league, and the 5th most against top competition. He's not just an up-and-coming star, he's already there - and he deserves the 2020 Norris Trophy.