What expected goals can tell us about the best lines in fantasy hockey

One of the reasons I like expected goals so much as an advanced statistic in hockey is because of the variation and depth of the models and forecasting that goes into it. As a weekend Excel warrior, watching the results of data scientists churn out similar, but different formulas to account for everything from angle of shots to time between rebounds is akin to a symphony on a spreadsheet! (Any attempt by myself would be a cacophony of cells or a tumultuous table.)

The slight variation on individual expected goals is evident when you surf through some of the more popular advanced metric websites. MoneyPuck.com has Auston Matthews tops in the league for expected goals, while EvolvingHockey.com lists Timo Meier at the top and NaturalStatTrick.com says its Zach Hyman. So there is variation there in how they run their models, but also predictability. All three of those players are among the top four on all three sites.

But the expected goals metric I like the most is MoneyPuck.com’s version that targets lines. When you pull the lens back to take a wider look at the on-ice outcomes for three players as opposed to one, logic dictates that you remove some of the margins for aberration that can occur when you get down to the individual level. (No, not errors in calculation; errors in the individual models’ predictability.)

There were 85 forward lines in the NHL that have played 60 minutes or more together at five-on-five as of Sunday. (I don’t know why, but I expected more than 2.7 lines per team to have stayed together for about six games. But I suppose injuries and the ongoing depth chart carousels are what keep us fantasy players on our toes.)

Let’s take a spin through some of those lines that have expected goals that are outpacing or lagging from their actual goal totals. As hinted, all the line data is from MoneyPuck.com through Saturday’s games and is at five-on-five, but we look at some individual expected goals stats from EvolvingHockey.com and NaturalStatTrick.com, too.

Lucky lines

Kirby Dach, Cole Caufield and Nick Suzuki, Montreal Canadiens (expected goals, 6.1; actual goals, 13): Suzuki might be the sole culprit here, as his 26.1% shot conversion overall is actually 28.0% at five-on-five and is going to be hard to maintain. There’s no denying the the trio, which only came together in the ninth game of the season on Oct. 29, has been getting a bit lucky with the goals.

There’s differing advice to hand out for all three players in a re-draft league: Hold or acquire Caufield, as he’s the shooter of the bunch; investigate selling high on Suzuki to see if there’s a market; pick up Dach as his overall 1.5 fantasy points per game (FPPG) belies his 1.8 FPPG since he joined the top line and he’s only rostered in 20.8% of leagues.

Jason Robertson, Joe Pavelski and Roope Hintz, Dallas Stars (expected goals, 12.5; actual goals, 19): You just knew this trio was on this list, as the 19 on-ice goals are just astounding at five-on-five through 21 games (and they added another one Monday).

It’s actually a pace that would top the incredible 72 goals at five-on-five by the former Calgary Flames top line from last season (which was easily the most goals by a line in all the MoneyPuck.com data going back to 2008-09).

But are they getting lucky or breaking the model? I wouldn’t let a fellow fantasy manager pry Robertson away for all the draft picks in the world, but he’s the individual here with the biggest gap between individual expected goals and his true output. The expected goals models have him between 5.8 and 6.0, but he’s actually potted 11 this season — he’s actually outperformed that metric ever since this line got together in 2020-21. I think Robertson will cool off a touch, but not enough to consider trying to sell him for anything less than a top-10 fantasy lock.

Andrei Kuzmenko, Elias Pettersson and Ilya Mikheyev, Vancouver Canucks (expected goals, 6.4; actual goals, 12): This became 7.0 expected goals and 14 actual goals later on Sunday after the data was calculated. So the run of luck continues for a trio that the Canucks found after a few games, broke up for a spell and then brought back together again. They are one of the leaders in the NHL in goals per 60 minutes (minimum 60 minutes at five-on-five), with only a rarely used Oilers combination doing better. Interestingly, it’s not Kuzmenko or Mikheyev driving the over-performance in goals, but Pettersson has an individual expected goals mark of between 4.4 and 4.8 compared to his 10 goals. He’s outpaced his individual expected goals in previous campaigns, but not by a margin like this. Expected a moderate cool-down period for Pettersson at some stage. On the other hand, Kuzmenko is still only rostered in 57.8 percent of leagues, but really should be universal. Mikheyev’s contributions aren’t enough for sustained fantasy relevance, but if this line sticks they probably will be at some point.

Other lucky lines:

• Alex Killorn, Nicholas Paul and Steven Stamkos, Tampa Bay Lightning (expected goals, 7.0; actual goals, 12).

• Josh Leivo, Brandon Saad and Ryan O’Reilly, St. Louis Blues (expected goals, 3.1; actual goals, 8).

• Anders Lee, Anthony Beauvillier and Brock Nelson, New York Islanders (expected goals, 8.4; actual goals, 13).

• Timo Meier, Tomas Hertl and Kevin Labanc, San Jose Sharks (expected goals, 8.9; actual goals, 13).

• Josh Bailey, Mathew Barzal and Oliver Wahlstrom, New York Islanders (expected goals, 3.2; actual goals, 7).

Troubled trios

Chris Kreider, Mika Zibanejad and Kaapo Kakko, New York Rangers (expected goals, 8.9; actual goals, 4): This trio hasn’t played together since Nov. 3, but the relevance here is that the Rangers are still putting their lines in a blender every other game, so they could (and maybe should) make a return. This is the line that the Rangers opened the season with at the top of the depth chart, but various losing streaks and dips in production from individuals have them still fiddling on a game-to-game basis.

This line did get 132 minutes together and actually ranks 11th in expected goals among all lines in the NHL, but the pucks just weren’t going in. Continued success on the power play has masked any of the five-on-five issues for Kreider and Zibanejad, so there isn’t really a buy-low opportunity here. That said, if I’m in a deeper league with lots of bench room, it may be worth checking to see if Kakko is available. He’s ranked between 36th and 25th in the league by the three individual expected goals models.

Michael Bunting, Auston Matthews and Mitchell Marner, Toronto Maple Leafs (expected goals, 8.9; actual goals, 5): This line is also no longer together, but in a way that matters a lot less for the individuals. The Leafs moved Marner next to John Tavares and replaced him on this line with William Nylander. The new trio is not doing as well as the original one in expected goals, but is doing much, much better in actual goals (and, for what it’s worth, in goals against).

In stark contrast to their world-beating performance as a trio last season, Bunting, Matthews and Marner were just not putting the puck in the net; they posted the best goals per 60 minutes rate in the NHL last season (minimum 200 minutes). But the swap of good players to play with other good players doesn’t have a huge individual impact here, other than the fresh lines seem to have sparked the offense at even strength. The trio of Bunting, Matthews and Nylander have already outpaced the original line for goals (well, technically they are tied now because the original trio had a freak eight seconds of ice time on Monday in which they scored, so, take that with a grain of salt).

What actionables are there here? Maybe Bunting is a buy-low candidate in some leagues. His overall 1.2 FPPG doesn’t look pretty, but he’s posted 1.8 FPPG since the switch to Nylander.

Brandon Hagel, Nikita Kucherov and Brayden Point, Tampa Bay Lightning (expected goals, 14.6; actual goals, 11): You would never know this trio is having “bad” luck based on their fantasy stats. They are tops in the league in expected goals though, so it’s a hard pace to keep up. They’ve been on the ice together for another seven goals on the power play, too, so that certainly keeps their fantasy profile in lockstep with the expectations.

The upshot here is that Hagel is absolutely, 100% for real and needs to be on your fantasy roster. This line has potential better fortunes ahead, his FPPG this season is 1.9, but he’s only only rosters in 20.9% of leagues. While Hagel isn’t guaranteed the ongoing power-play minutes, as we’ve seen him replaced by Alex Killorn at times, this even-strength trio is not likely to be the one at risk of a breakup with the looming return of Anthony Cirelli.

Other troubled trios:

• Artturi Lehkonen, Mikko Rantanen and Nathan MacKinnon, Colorado Avalanche (expected goals, 9.9; actual goals, 8).

• Gabriel Vilardi, Adrian Kempe and Anze Kopitar, Los Angeles Kings (expected goals, 7.0; actual goals, 4).

• Brad Marchand, Patrice Bergeron and Jake DeBrusk, Boston Bruins (expected goals, 4.9; actual goals, 2).

• Jonathan Huberdeau, Elias Lindholm and Tyler Toffoli, Calgary Flames (expected goals, 3.7; actual goals, 1).

• Trevor Moore, Phillip Danault and Viktor Arvidsson, Los Angeles Kings (expected goals, 10.7; actual goals, 8).