Connor McDavid was brilliant as a rookie in 2015-16 and anybody paying attention knows it.

What many don’t realize is how much his linemates mattered, and which linemates he actually clicked with. The conventional wisdom about which players he had success with in 2015-16 is almost entirely wrong.

Previously in this series:

As before, this chart was generated via Puckalytics’ SuperWOWY function, and I have shown the forward combinations for each line on the far left, followed by 5-on-5 minutes together, goals and Corsi plus/minus as a unit. I’ve left out zone starts this time, just because it makes the chart too big, but will be noting them in the write-up below.

For the sake of reference, I’ve also included generic first, second, third and fourth lines. Again using Puckalytics, I ranked the top 360 forwards in the league by ice-time. The averages above are the unweighted totals for the NHL’s 90 most-used forwards (three per team), followed by the next 90 and so on. These are back-of-envelope calculations but they do give us a point of comparison.

There’s no question as to where McDavid was most effective: It was between Benoit Pouliot and Jordan Eberle.

Zone starts aren’t included here, but this unit actually started more shifts in the defensive than offensive zone—odd for a scoring line—and had a brilliant 56 percent Corsi. The goal numbers were slightly worse, at 53 percent, but given that we’re only talking about ~two hours and one fewer goal against would have bumped this trio up to 56 percent, I wouldn’t worry about it. Pouliot-McDavid-Eberle is a fantastic line, even compared to other NHL top lines and even without adjusting for zonestarts and playing with Edmonton’s defence.

Switch Eberle for Yakupov, and things get worse. The goals are actually a little better in this small sample, but the Corsi number drops by more than four percent. There’s some important usage context, too. With Eberle, this line started 57 shifts in the defensive zone and 51 in the offensive zone (six more d-zone draws). With Yakupov, this line started 20 shifts in the defensive zone and 47 in the offensive zone (27 more o-zone draws). We see a similar drop-off when Yakupov is substituted for Pouliot.

The line at the very bottom of this list is going to surprise some people. Patrick Maroon s widely perceived as a strong complement to McDavid because of his late-season performance on the line. However, the only time that we see McDavid’s Corsi fall below 50 percent is when he played with Maroon.

This line did a good job of scoring goals, but we’re only talking about two hours of ice-time together; over such a small sample, I trust the Corsi number a lot more than I do the goals number (256 total events vs. 13 total events). The caveat here is that McLellan used this line a lot in the defensive zone, but not enough to justify the unit’s relatively poor possession work.

Even for a player as skilled as McDavid, linemates matter, and a lot of what has been said about his linemates is wrong.

The one bit of conventional wisdom that is correct is that Eberle and McDavid worked well together. It wasn’t a bullet-proof arrangement—with a weaker left wing, the line struggled—but as a duo these players were good together.

My other three takeaways run directly against conventional wisdom.

Pouliot’s work on the line simply doesn’t get nearly enough credit. The drop-off from Pouliot to other left wings was almost exactly the same as the drop-off from Eberle to other right wings. If we’re looking at duos, Pouliot-McDavid is bigger and cheaper than McDavid-Eberle and if anything was slightly better at hockey in 2015-16. Given the number of complementary right wings in free agency this year and the relative value of the two players, Edmonton’s probably better off keeping Pouliot and trading Eberle than it would be keeping Eberle and trading Pouliot.

I’d be inclined to keep both players, for what it’s worth. Pouliot-McDavid-Eberle was a brilliant line and there’s no reason to move away from it.

Maroon was not a particularly good fit. Doubtless he’ll see some time with McDavid next season, but if the Oilers have him written in ink as their second line left wing the team is setting itself up to fail. If McDavid’s line runs a 48 percent Corsi next year, the goal numbers will regress and it’s hard to imagine Edmonton making the playoffs.

Yakupov was also not a particularly good fit. McDavid brought out the best in him, but he didn’t bring out the best in McDavid. Additionally, based on zone starts it is obvious that Todd McLellan stops trusting the line as soon as Yakupov was placed on it.

The biggest takeaway though relates to the overall team. In the Hall piece I noted that Hall/Nugent-Hopkins or Hall/Draisaitl could “reasonably be regarded as the foundation of a competent first line or a brilliant second unit.” Combine that with a Pouliot-McDavid-Eberle first line and Edmonton’s top-six forward group should be better than the vast majority of NHL teams.