The NHL playoffs mean the script gets flipped on a lot of things. The intensity of the games and the brief window to succeed means shooting and save percentage become paramount. The hot goalie and the shooter who’s hitting the net are kings. The guy with no goals gets roasted on every intermission panel.

But there’s another theme that gets reversed. Fans and media who spend the regular season overemphasizing the effects of Quality of Competition, spend the playoffs underemphasizing it.

This caught my eye today:

Now, Sean Tierney doesn’t really mean “best” in the sense of an overall measure of value. He means, I hope, “best” in the sense of these results devoid of context. Dom Luszczyszyn provides the context. Minor quibble: I would call the terrifying trio of the Colorado Avalanche the best line in hockey.

I’m going to use some Natural Stat Trick stats here to get a handle on these numbers. Ennis has played two games and has a laughably huge Relative Expected Goals Percentage at score-adjusted five-on-five. So we’re seeing big effects of variance and QOC in a very small sample of 13:22.

Frederik Gauthier and Trevor Moore at about 28 minutes each are good, but slightly less stratospherically excellent. Connor Brown, who is only just above average in Corsi, leads the team by this measure too.

The worst players by Relative Corsi are: John Tavares, Andreas Johnsson, Mitch Marner, Kasperi Kapanen, Ron Hainsey and Jake Muzzin. Nikita Zaitsev and Zach Hyman, the other members of the five-man unit that faces the Bergeron line are much better, just under team average.

But the Rel xGF% of these players is very different. In small samples, there are big effects. And shot quality can add a much bigger skew to the shot share over the 60-70 minutes these players have been on the ice than it will when “washed out over a regular season”, to use the popular phrase. Nothing is being washed out here.

The worst players by Rel xGF% are: Andreas Johnsson, Travis Dermott, Kasperi Kapanen, Auston Matthews, Jake Muzzin, Ron Hainsey and John Tavares. Then there is a whole bunch of players until we get to Marner, Zaitsev and Hyman, who are well above team average.

I’m sorry to tell you all, but so far Zaitsev is outperforming Muzzin. And Marner and Hyman are not merely above water, they are excellent.

Now all of these players play some of their minutes against other lines than Bergeron. And what do small samples have? Big effects, that’s correct. When all three are on the ice together, Natural Stat Trick tells me, they are well above team average in relative Expected Goals, but just below in shot share.

Much more importantly when I look at this game-by-game, I see this:

Game One: Leafs win: Tavares line is at 21.39 Rel xGF%
Game Two: Leafs lose: Tavares line is at -38.74
Game Three: Leafs win: Tavares line is at 7.80
Game Four: Leafs lose: Tavares line is at -19.80

John Tavares has a 2-1 ratio of time vs Bergeron and time vs other lines. And against Bergeron, his Corsi For Percentage is 42, but his Expected Goals Percentage is 50. Tavares’s time vs the Krejci or Coyle lines is gravy. He’s dominating them. And so it seems, is the new-look third line with William Nylander at centre.

That shiny fourth line that everyone finds so fun to watch has played about two minutes of total time vs anyone other than the Bruins fourth line, who I think we can conclude aren’t very good. Playing them more won’t help their stats or the Leafs chances.

The truth is, the Leafs rise and fall on the success of John Tavares, Zach Hyman, Mitch Marner, Jake Muzzin and Nikita Zaitsev at shutting down the Bergeron line enough to give themselves some chances against their weak defending, and the Matthews and Nylander lines some chances against the very ineffective rest of the Bruins.

So, no, the “best” line is not the fourth line. The best line is helmed by John Tavares, and if the Leafs win this series, it’s down to him.