Nothing has enthralled the masses of Leafs fans over the years since 2016 more than the ice time of Auston Matthews. When he was young, you could stir people up by noting if he didn’t play more than John Tavares every game — even if it was only by a few seconds. When he got older, and Boston became the playoff opponent, it was harder to find cases that persisted for a full game, but the idea had stuck that he was underplayed.

The easy regular season way to froth things up was to pick an early game full of penalties taken by the Leafs and use all-situations ice time. Some grinder PK forward (it was Leo Komarov one time) will have much more time on ice than Matthews, and it’s time to be angry online at the coach.

This was all a bit surreal because NHL standard practice had already undergone a big change by the time Matthews was surpassing Tavares and taking the number one spot. Ryan O’Rielly, playing all game states on the Buffalo Sabres, was held up as an example of a dying breed — the forward who played defenceman minutes every night. Even he doesn’t do that anymore, but in 2018, complaints that Matthews wasn’t playing enough seemed to run counter to the understanding of optimal shift length and player usage. Then the coach changed.

Sheldon Keefe liked Auston Matthews a lot (as does ever other coach he’s had) and he played him big minutes. Really big minutes. And suddenly the grievance flipped to concerns he was playing him too much, or his shifts were too long or both. It became a truthy truism that in 2019-2020, Auston Matthews was on the ice too much.

There is no actual evidence if specific players should or should not play a given amount. This is a lot like goalie rest, in that it’s so hard to judge the physical state of a player even if you are their doctors. Fans are free to make up stories about what should happen with a player. Whatever the coach you don’t like does is wrong, is the main story. If the team is losing, everything is wrong, if it wins, everything is right. None of which is anything more than a lot of post hoc narratives.

Now with Sheldon Keefe experimenting with a “balanced lineup” after spending much of the season with a third line that was seen as a shutdown line, the issue of ice time is going to be front and centre again.

The argument in favour of the third line having a player as good as William Nylander on it is that you can generate offence more effectively with three lines by taken advantage of the other team’s weaker lines in matchups and continue to barely play the fourth line. The result is you prosper with more goals. The trick is you need enough forwards of the right type to pull this off.

The argument against this plan is that in order to use the third line effectively, you have to play that line more than you have been. You can’t play the fourth line negative minutes and bank game time. You have to take those minutes from somewhere else or else play that scoring line shutdown line minutes. A top-heavy team that relies on one line to get all their goals is a problem on the nights that line doesn’t work — like say, game seven of a recent playoff series. However, if the cure is playing one of the best players in the NHL less, is that worse than the disease?

I used to be in favour of a “roll three lines” approach, back when the Leafs seemed to be awash is scoring forwards and Matthews was still a bit green in some ways. Now the reality is different. There is a dearth of skill, and any configuration that takes minutes away from a top forward is a mistake. Taking Matthews minutes away seems like a pointless mistake. I see the reason for the impulse to mix it up, and it’s not about Matthews. It’s about Nylander and Tavares getting caved in and not scoring while the third line doesn’t get caved in and can’t score.

I see the actual problem as the lack of skilled wingers on the left side who are legitimate scoring line players, so it seems clever scheme-y to try to solve that by making three scoring lines. To date there has been a rotating usage that results in four forwards by time on ice at five-on-five consisting of Mattews, Marner, Tavares, Nylander, or two/thirds each of the top two lines. The fifth player, Michael Bunting, is almost a full minute less per game than Nylander. It’s obvious that some creative usage has already been employed, just not to the extent that you see it in line rushes in practice. Matthews plays the most of any forward, and that should always be true until the Leafs trade for Connor McDavid, but he’s lagging behind the usual league leaders.

Connor McDavid is the usual first or second player by time on ice per game. Matthews, in the two seasons prior to this one, has been a few hundredths of seconds behind him. This season so far, Matthews is over half a minute back, and is playing similar minutes to Patrick Kane and Nathan MacKinnon. This makes me uneasy about trying to balance the lineup and potentially playing him even less.

I can’t tell you if Auston Matthews’ results degrade if he plays more than a specific number of minutes, so I don’t know his magic correct number. I’m not here to create narratives out of thin air about injury recovery (although his surgery is very common and not a big deal). I’m not trying to imply anything about how good or bad a particular amount of ice time is. All of that is guessing. All I’ll say is that playing Auston Matthews even less and giving those minutes to David Kämpf is how the Islanders would do things, and isn’t going to lead to winning more games.

How Matthews got to slightly less minutes is interesting. In the recap of the game vs the Islanders, I said (guessing) that Matthews is making up his ice time over his usual linemates by playing with the fourth line in little sneaky shifts where suddenly there he is. This isn’t new, per se. Matthews has about 20 minutes with Frederick Gauthier in his one season of glory in the NHL. But it seems to have increased.

I thought I should stop guessing and look into this, so using the shift query tool at Evolving Hockey I compared Auston Matthews shift lengths for all of the last three seasons where Sheldon Keefe has been the coach. That means the 2019-2020 season has a few games chopped off the front. This is regular season only.

Auston Matthews shift length averages

SeasonPeriod 1Period 2Period 3 OvertimeOverall

The by-period splits are interesting, in that he’s the man in the second period. Hold the lead? Get the lead? Doesn’t matter, he’s the guy. Overtime shifts are fairly volatile because there’s so few and situational in nature, so I’d mostly ignore that, but the overall averages are where a subtle change can be seen.

Why, though? To answer that fully would require mind reading, but some of this might be the Leafs having so many leads late into the game so that Keefe feels less inclined to shorten the bench even more than it usually is.  A deeper dive than the averages revealed one clue as to what’s changed, though.

I looked at the distribution of shift lengths, and Matthews’ are broadly similar each season, but this year has one unusual aspect to it: The number of shifts between 30 and 35 seconds is already the highest it’s been in these three seasons. He does not have a lot of really long shifts in any year, and most are in the 40 to 65 second range in all three seasons, the same as most top-six forwards in the NHL. Matthews might have a few more over the one minute mark than most players, but the increase in short ones seems to imply that his shifts with the depth players are increasing. This is a subtle change, and the proportion of total shifts below 35 seconds is at 17%, the highest in the three years, up from 16% in 1920 and 15.5% last season.

My guess looks like it was fairly accurate. The amount of analysis of the shifts with other players necessary to figure this out for sure is complex, and won’t make any of us smarter. But Matthews already has 42 minutes with Wayne Simmonds. And while some of that is Simmonds playing some top line wing, a lot of it isn’t. Matthews has over 18 minutes with Pierre Engvall and a total of at least 10 minutes with confirmed fourth liners like Kyle Clifford and Brett Seney. He’s getting around a lot this year. It should be remembered that five-on-five shifts with Tavares aren’t new or unusual, and they have some time together in every season. The lineup card is a guideline, not reality, and the choices a coach makes in-game are based on context.

Some of the context for this season to date is that it’s the first half where the Leafs have won a lot and the urge to go for the big gun isn’t as strong. So, assigning reasons to Keefe’s choices is likely going to be more about our opinion of Keefe than reality.

So what’s the difference between outright playing Matthews on a line without Marner or Nylander and just shoving him onto other lines randomly in-game? Likely not much, as long as he doesn’t play less while he’s carrying lesser players. The real question about the current concept is what’s the point of William Nylander playing with David Kämpf and Alex Kerfoot?

Maybe I should have looked at Nylander’s shifts over the years.