Mitch Marner and his Primary Assist Magic

You may have heard by now that Mitch Marner remains unsigned by the Toronto Maple Leafs as we begin to approach the opening of training camps for the 2019-20 NHL season. There seems to be a gap between how the Marner camp views his value compared to how the Leafs think he should be paid. Marner put up fantastic numbers last year, picking up 94 points and finishing 4th in the NHL with 68 assists. His results were even more impressive at 5v5, where Marner led the NHL in primary assists with 36, five more than Sidney Crosby. Unlike secondary assists, which are mostly luck, primary points are an important, repeatable skill, so Marner putting up the best total in the NHL at 5v5 is quite an accomplishment.

This would seem to set the stage for Marner to get a huge payday. After all, if he can put up the highest rate of primary assists in the NHL then he has a very strong claim to a big salary. However, if you dig into the numbers a bit more deeply, I think there are reasons to be concerned that Marner isn’t going to be able to replicate his fantastic result next season.

One problem with looking at the league leaders in counting statistics like goals and assists is that they’re actually a combination of two different things: player skill and ice time. If two players are equally talented at scoring goals, but one gets more ice time, then the player who gets more ice time will finish with more total goals even though they aren’t actually better than the player with less ice time. One way to account for this is to use rate statistics; the most common way to do this is to state how good a player was per 60 minutes of ice time they received.

As it happens, using rate statistics makes Mitch Marner’s playmaking ability look even better than using his raw assist total. Among players with at least 500 minutes of 5v5 ice time, Marner had by far the best rate of primary assists last year. Here are the top five players in terms of 5v5 primary assist rate last season:

Marner is not just in first place, he’s in first place by a very wide margin. Only four players were even within half an assist per 60 minutes of ice time. That’s impressive enough, but it gets better: Mitch Marner’s 1.78 primary assists per 60 minutes is the highest rate of any player in the entire “analytics” era (since 2007 when the NHL expanded its real-time data collection) with at least 50 games played in a season.

At this point, you’re surely wondering when I’ll get to the bit that explains why this isn’t as good as it seems. I think it’s time to dive into that question.

So Mitch Marner had a historically good season last year. But how good, exactly, was it? And why do these historical comparisons bode ill for Toronto’s energetic young winger? First, here’s a list of the best seasons between 2007 and 2017, a list that comprises all 25 player seasons with at least 1.25 A1/60 at 5v5:

You may notice that this list doesn’t include any players from the 2011-12, 2012-13, or 2018-19 seasons. There’s a reason for that that I’ll get to in a moment.

So, in over a decade there’s only been one player who put up a primary assist rate comparable to Marner last season – Daniel Sedin in 2009-10. Again, you can see how much of an outlier Marner’s results were. You might interpret that to mean that he’s a generational play-maker, and while I can not deny that that’s one possibility, I think there’s another that’s more likely.

Let’s look at that last list again, but this time let’s see what each player’s primary assist rate was the following season. (This is the reason for omitting the three seasons listed above – none of them had a subsequent 82-game season to compare with.)

One thing that immediately jumps out is that none of these players saw their primary assist rate go up the next season. Every player who has put up at least 1.25 primary assists per hour at 5v5 has seen their A1 rate fall the next season. Only two players managed to remain above the 1.25 A1/60 mark in consecutive seasons, and they’re both named Sedin. Henrik, remarkably, nearly did it three seasons in a row from 2009-10 to 2011-12 (1.58, 1.50, 1.24). No one else has managed to do it in over a decade; not Sidney Crosby, not Connor McDavid, not Joe Thornton.

Now, there’s no way for me to conclusively prove that Mitch Marner is going to meet the same fate as all of these other players. Maybe a year or two from now I’ll be able to update this post and talk about how Marner is the most successful passer at even strength since Henrik Sedin. But I think on balance a more likely outcome is that Marner will see his rate of primary assists fall pretty considerably. It’s happened to all the best passers in the game (except Henrik Sedin, who is eternal), so it’s likely to happen to Marner too.

As of the time I’m writing this (Sunday, September 8) Marner and the Toronto Maple Leafs remain at a stand-off over what kind of term and dollar value his next contract should have. I think what I’ve pointed out above has pretty clear implications for the negotiations. If Marner really is a generational play-maker, then he’s got a pretty good case to be asking to be paid like one of the best players in the league. But if there was some luck in his results last season, then the Leafs have good incentive to hold the line and try to avoid overpaying him based on results that will be difficult to replicate.

One thing that makes me sceptical of Marner’s results last year are his results in previous seasons. Here is how well Marner has done with each of the centres he’s played at least 200 minutes within his career so far:

Marner’s results last year were a shocking increase over his previous results. The best he had done with any centre prior to being paired up with John Tavares was only half as good as he did with Tavares last year. I think there’s a pretty good case that playing with an elite goal scorer for a long stretch of time, a player who has a reputation for boosting his linemates’ totals, helped bump Marner’s results up. (I also think Marner boosted Tavares; it’s not a one-way street.) They may have also gotten a bit lucky, as their on-ice S% of 11.3% is pretty high. At any rate, based on Marner’s previous history, I think there’s good reason to believe his primary assist rate last year is going to be difficult to maintain moving forward.

In the list of players above, all 25 of them saw some regression from one season to the next. But how much? On average they saw a drop of 0.44 A1/60 the next season, with a nearly identical median of 0.45. If we adjust Marner’s results last year down by 0.44 A1/60, he falls from 36 primary assists to 27. It should be noted that 27 primary assists at 5v5 is still really good – it would have ranked 4th best in the NHL last season behind Sidney Crosby, Artemi Panarin, and Connor McDavid, slightly ahead of Nikita Kucherov. But it would drop Marner to 85 total points, a bit behind other top restricted free agents like Mikko Rantanen and Brayden Point.

If Marner’s true talent is closer to 80 points than 100, I think that changes the negotiation.

At that point, we’re talking about a player who doesn’t have a very good case to be making more money than any of the other top RFAs. On the other hand, maybe Marner really is just an exceptionally gifted player who will beat the regression that’s pulled down the numbers of virtually every other top passer in the game. It’s an important question for the Leafs to be able to answer with so much money and cap space on the line.