If I had a Hart Trophy ballot and had to cast it today, my top two picks would be Mark Stone and Sidney Crosby. I might consider a third place vote for Nikita Kucherov as a nod to his historical scoring accomplishment, but would probably go with Aleksander Barkov or Brayden Point instead.
I suspect this opinion will put me in the minority, and when the ballots are counted, Kuch will win his first Hart Trophy. That’s a fine outcome. He’s played well enough to be in the conversation. But how we have that conversation is changing. And in a few years, the way we talk about determining the most valuable player might be entirely different than it is now.
When I started writing this article, I thought I’d write something different — just an article about whether or not Kucherov should win the Hart. But that’s not what I wrote. I wrote about about the way we evaluate players and about the growing pains that come with it.
Let’s start with the fun part.
Nikita Kucherov is having the best scoring season in recent history
Kucherov has scored 128 points this season. That’s the most since 1995-1996 when Mario Lemieux led the league with 161. Kucherov’s total surpasses Joe Thornton’s salary cap era record of 125 set in 2005-2006 and is the first 120 point season since Sidney Crosby in 2006-2007. In Crosby’s 120 point season, he was the obvious Hart winner. The previous season, Thornton and Jagr both broke the 120 point threshold. They came first and second respectively for the Hart Trophy dominating the voting.
In 06-07, Crosby won the Hart Trophy comfortably. He got 91 of the 143 first place votes. That season, he outscored Thornton, who was second in points, by six. Kucherov put twelve points between himself and the player in second, who happens to be Connor McDavid.
In the salary cap era, if a player scores 120+ points, they win the Hart unless another player outscores them in the same season. The last player to score 120 without winning the Hart was Jagr in 2000-2001 when he scored 121 and lost to Joe Sakic who totaled 118.
Viewed through this historical lens, Kucherov seems like a lock to win the Hart. He’s scoring like we haven’t seen in over two decades. And if we look at his scoring in terms of points per minute played instead of raw totals, his numbers get outright absurd. By that measure, he’s setting a new record among players who played a full season for as long as we have data, which is back to 1998-1999.
The evolution of the stats we use, and what we can learn from baseball
Things have changed since Crosby’s Hart winning season in 2006-2007. Starting in 2007-2008, the NHL provided more data. We’re no longer limited only to counting stats like points to make decisions on who is the best player during a given season.
That’s a good thing. Better data means better metrics to evaluate players. And better metrics means we have an opportunity to make better decisions. The “we” in this case is everyone involved in hockey. If teams can leverage that data, they can make better decisions about which players to give more ice time and what to offer during contract negotiations. If the public can leverage that data, they can make better decisions about how to talk about the game and which players to celebrate.
But better data and better metrics are also a complicated thing. It means we’ll learn that some of the things we always thought were wrong. It means we’ll learn that some of the benchmarks we relied on to identify greatness weren’t as reliable as we thought. It means changing the way we make decisions and therefore, acknowledging that some of our previous decisions were misguided.
When I think about the transition hockey is making into an era with more robust statistics, I look to baseball for help. Not because the games are similar. They aren’t. But because baseball has grappled with these concepts for longer. And even if the games are dissimilar, the process of figuring out how to incorporate new information and its impact on the game’s future and its history is comparable.
Baseball has deprecated statistics that 30 years ago, were considered vital. Pitcher wins are now considered mostly meaningless. Felix Hernandez won the Cy Young award as the best pitcher in 2010 despite a win-loss record of 13-12. Even 20 years ago, considering him for the award with that record would have been laughable.
But baseball has gotten smarter and realized that pitchers have little control over whether they win or not. All they can do is pitch well. Whether their defense plays well or the offense scores runs is out of their scope. And in 2010, Hernandez was the best pitcher. The Mariners’ offense forsook him but that wasn’t his fault. So he rightly won the award. And pitcher wins were buried in the graveyard of things we used to think were important.
Hockey has not yet had a moment like this. Some statistics are losing favor. Goalie wins are now understood to be silly for largely the same reasons as pitcher wins. Goals Against Average is also nearly extinct as we’ve learned that number is dependent on the number of shots the defense allows, which the goalie can’t control.
For skaters, plus-minus is dying a slow painful death. Every citation now comes with a caveat. Either of its inherent weakness or, for the particularly stubborn, a condescending shot at the new way of thinking. Eventually, the first type of caveat will disappear as those using it realize they have better tools available to make their point. The second type of caveat will also disappear as those making them will be wrong so frequently that they become irrelevant.
Points, WAR and transitioning to better tools
But points as a statistic are not dead. They aren’t even dying yet. The criticism of them is just starting to mount. It comes from exactly the places one would expect. The people at the forefront of statistical innovation in hockey are looking at the results of their hard work and rightfully wondering why anyone would use something so devoid of context as points to decide who the best players are.
Points suffer from some of the same problems as the stats we discussed above. They are inherently dependent on a player’s teammates. If two players, one on the Lightning and one on the Senators, make the exact same plays over the course of a season, the one on the Lightning will have more points. A great pass is a great pass. But a great pass to Steven Stamkos does not have the same likelihood of becoming a goal as a great pass to Bobby Ryan.
Accounting for things like quality of teammates, quality of competition, and usage means more complicated math. In the past, we’ve done this in our brains. We looked at point totals or plus-minus or even shot metrics, and did some adjusting according to our perception of the context in which those results occurred. Now, we have models that do this explicitly.
The most prominent approach is Wins Above Replacement (WAR). The only publicly available WAR model currently is the one from Evolving Hockey. Fortunately, it’s a very good WAR model.
Baseball has long had WAR models. Hockey is just getting started. Baseball had a distinct advantage in that their WAR models could be applied historically so they could look at over a hundred years of history. That gave the models credence. Mostly, their outputs agreed with historical perceptions. Some players, like Tim Raines, suddenly looked much better than they got credit for during their career. Other players, like Jack Morris, looked a bit worse, the product more of longevity and being on good teams than was obvious when they were playing.
We can’t go through this exercise in hockey. The data is so sparse prior to 2007-2008 that constructing a good WAR model is impossible. That deprives us of the ability for a WAR model to confirm how great Wayne Gretzky and Bobby Orr were. It removes the opportunity to look at a hundred years of history and say, “See! We mostly agree!”
Instead, we’re forced to look at a condensed data set with barely over a decade of history. The lack of a big picture forces the conversation into the small picture. And in a small picture, trends are less obvious. Consensus building is more difficult. Minor disagreements are magnified.
Back to Kucherov and this year’s Hart Trophy race
This year will be an example where one of those minor disagreements is magnified. Nikita Kucherov scored 128 points. That’s first in the NHL. Nikita Kucherov has 3.2 WAR. That’s 16th in the NHL. By points or by WAR, he’s one of the best players in the league. But that’s not precise enough for Hart Trophy voting. For that, we need to know who is THE best. And in points, he is. In WAR, he isn’t.
WAR is a better statistic for holistic player evaluation than points. Let’s get that out of the way. Arguing that points are a better statistic is arguing for a flat Earth. And in that sense, maybe we should just look at the WAR leaderboard, vote the top-three players for Hart, and move on to the next question.
But WAR, like everything, is imperfect. While the numbers on a leaderboard might say “4.4” or “3.2”, the metric is not that precise in reality. Model outputs have error attached to them. The Evolving Hockey WAR doesn’t have explicit uncertainty measures but its creators have suggested that players within a half win of one another can be considered to be playing similarly well. Previous models have put that number at a full win.
In that context, maybe considering players within a half win, or up to a full win in certain circumstances, of the leader is a sensible approach. For this season, that would include eleven players — not sixteen. Meaning that Nikita Kucherov is currently outside the window where someone using the best player evaluation tool available could reasonably argue he’s been the best player this season.
This is an unpopular opinion, especially on a blog like this. Unpopular with readers. Unpopular with the other writers. And unpopular with me personally.
Arguing for Kucherov to win the Hart means trying to find something else to bump him up the Trophy ranks. Some of the old approaches would work well here. The Lightning have the best record in the NHL by a wide margin. So give him some credit for team results. He’s been one of the best players in the league for five years now and hasn’t won it yet. He’s due and he’s had a great season.
An argument that I have some time for and might even make if I had less tolerance for bad faith is the historic one: 120 points doesn’t happen often anymore. And if a player does that, they win the trophy unless someone else scores more or their advanced metrics are so bad as to make it impossible to justify. Maybe I’m too sentimental but I do see value in historical continuity and in the idea that certain numbers hold important places in the game. They’re thresholds. If you cross them, you get a reward.
But the problem with that is that sometimes, we’re wrong about how important certain numbers are. And to continue to agree to those thresholds is to continue to be wrong.
None of this matters this year
This season, I think we might all agree to just be wrong. I expect Nikita Kucherov to win the Hart. And in the annals of unjust NHL trophy decisions, this would hardly be a blip. So he’s 1.2 WAR behind the leaders. He’s still a top twenty player in the league this season. Hardly an egregious injustice.
But five years from now, in the same set of circumstances, I’m not sure he would win. As new statistics like WAR gain staying power and impact the way we think about players, legacy statistics like points will fade in importance.
Today, writing this article saying saying a player with 120 points shouldn’t win the Hart because he has 1.2 less WAR than the leaders will be panned, and I’m expecting it. But most of the criticism will be grounded in resistance to change as opposed to reason. The stones thrown at WAR will shatter the glass walls of points, shots, and the other statistics used to construct the counter arguments.
Every attack that can be levied at WAR can be directed right back at whatever alternative is being suggested. The reason people look to other statistics before WAR as a player evaluation tool is because those stats came first. And because they came first, we’ve formed lots of opinions based on them. Changing those opinions can be hard, so instead we resist the new thing and defend the old thing.
But eventually change happens. In five years, I suspect we’ll have moved toward understanding that points were always a flawed stat and that while a player’s scoring is one of the few things we can use to compare them historically, it shouldn’t be the thing we use to compare them to their contemporaries.