The NHL’s annual All-Star weekend is upon us, and that means fast skaters, blistering shots, mascot hijinks and even a new trick-shot competition. For the fifth-straight year, it also means a heavy focus on the 3-on-3 format, including a women’s All-Star Game on Friday night and the four-team men’s tournament on Saturday. The NHL seldom spent any time playing at 3-on-3 before 2015-16, when the league changed its overtime rules to play the tie-breaking period with three skaters per side. But ever since, 3-on-3 hockey has become a somewhat regular fact of NHL life:
The increasing popularity of 3-on-3 makes a lot of sense. With fewer skaters on the ice to clog up the same amount of territory, offensive players have a lot more time to think and room to move — and consequently, scoring goes through the roof. According to data from the statistics site Corsica Hockey, 5.97 goals have been scored per 60 minutes of 3-on-3 hockey in the NHL this season, a whopping 140 percent increase over the scoring rate (2.49 goals per 60 minutes) of regular 5-on-5 play. It’s been bad for goalies, who not only face 6.6 more shots per 60 minutes but also find themselves allowing around twice as many goals per shot at 3-on-3 (16.1 percent) than at 5-on-5 (8.2 percent). But it’s good news for both the league, which wants more games decided before the shootout,<a class="espn-footnote-link" data-footnote-id="1" href="https://fivethirtyeight.com/features/the-nhls-3-on-3-all-star-game-is-a-scorers-delight/#fn-1" data-footnote-content="
Mission accomplished: Since the 2015-16 season, 64.9 percent of games that went past regulation ended before going to a shootout, compared with 43.5 percent in the five seasons before the league switched to 3-on-3 OT.
“>1 and for those players skilled enough to best exploit the extra space on the ice.
That’s what makes 3-on-3 an especially ideal format for an All-Star Game, which is designed to assemble the best — or at least the most offensively productive — players in the sport. And indeed, the most prolific 3-on-3 scorer of the past three seasons, Edmonton’s Leon Draisaitl, will be representing the Pacific Division this weekend. But the best offensive threats in “normal” hockey aren’t always the top 3-on-3 players. For every current All-Star with at least 3,000 minutes at 5-on-5 and 25 minutes at 3-on-3 since the 2017-18 season, I mapped out their Goals Created (GC) per 60 minutes in each situation, looking for the players whose production increased (or decreased) the most with three skaters on each side.
On the high side, Draisaitl has been one of the biggest gainers at 3-on-3, elevating his Goals Created per 60 minutes by 2.91 at 3-on-3. That change ranked fifth among All-Stars, behind St. Louis’s David Perron (+3.89), Winnipeg’s Mark Scheifele (+3.80), Dallas’s Tyler Seguin (+3.35) and Calgary superpest Matthew Tkachuk (+3.16), just ahead of Columbus defenseman Seth Jones (+2.82). Most of those guys are at least solid offensive producers overall — Draisaitl has actually been historic this season — but Jones is a great outlier toward the left side of the chart. More of an all-around blueliner than a pure offensive defenseman, Jones doesn’t score very many points at 5-on-5. At 3-on-3, however, he has 4 goals and 4 assists in 62.4 minutes of play, an especially huge scoring rate coming from the back line. (No other All-Star defenseman has created more than 2 goals per 60 at 3-on-3 over the past three years.)
Meanwhile, there are some stars who do surprisingly little damage at 3-on-3, despite the extra chance for creativity. Boston Bruins right wing David Pastrnak currently leads the NHL in goals with 37 markers in just 51 games so far this season, one of the best goals-per-game rates in recent history. By any measure, Pastrnak is an elite offensive talent, but he has produced only 2 points (and zero goals) in 50.3 minutes of 3-on-3 action over the past three seasons — good for a Goals Created rate 0.40 per 60 minutes lower than during 5-on-5 play. The only All-Star whose production dropped off more at 3-on-3 was Minnesota’s Eric Staal (-0.84), who didn’t score a single point at 3-on-3 in 33.5 minutes. Pastrnak, Staal and Chris Kreider of the Rangers (-0.26) were the only three All-Stars whose stats somehow got worse at 3-on-3 than at 5-on-5. That’s not supposed to happen to players with All-Star skills, although it’s also a testament to the fact that anything can happen in small samples and weird man-strength situations.
And then there are the goalies, who were already tasked with an impossible job under pre-2016 All-Star conditions, when per-game scoring increased by 236 percent compared with the regular season. Then things got even worse between the pipes when the 3-on-3 format was adopted. So no All-Star netminder has had a better save percentage at 3-on-3 than at 5-on-5 — though some dropped off less than others.
Tristan Jarry, who is having a breakout season for the Pittsburgh Penguins, checks in at No. 1, though he has faced a paltry 17 shots at 3-on-3. (For a little context, it generally takes 3,000 shots for a goalie’s long-term performance to stabilize, so we’re dealing with absolutely microscopic samples here.) As alternative options atop the leaderboard, Connor Hellebuyck of the Jets and Toronto’s Frederik Andersen have stared down more attempts at 3-on-3 and stopped almost as many of them relative to their 5-on-5 numbers. But Washington’s Braden Holtby is a huge outlier in the opposite direction. Usually one of the top netminders in the game (although he has struggled this season), Holtby has let in a wretched 22.2 percent of opposing shots during 3-on-3 play, easily making him the worst All-Star goalie in those situations.
The Central Division team, led by Colorado’s Nathan MacKinnon and backstopped by Hellebuyck, probably has the best on-paper combination of offensive talent, goaltending and collective 3-on-3 track record. But whoever wins the All-Star tournament, fans are in store for a lot of fast-paced, high-scoring action — thanks in large part to the fundamental nature of the 3-on-3 format. (Sorry, goalies.)