Islanders 5v5 Success: The Goaltending or Team Defense?

After 56 games it is clear that the New York Islanders have been one of the best teams in the NHL in terms of preventing goals. In fact their 5v5 goals-against rate (per-60) of 1.83 leads the NHL, while the Boston Bruins are the only other team below 2.00, at 1.91.

Is there a reasonable way to tell if the Islanders’ success preventing goals at 5v5 has been more driven by goaltending or by team-defense? We’ll look at one method today: comparing expected goals allowed to actual goals allowed, at 5v5.

We will break the following chart into five sections— five parts of the season— to help illustrate how expected goals against (blue) and actual goals against (red) compare to the league-average for goals allowed per-60 (green line). Keep in mind that points on the blue and red lines represent a 10-game rolling average.

One key turning point this season: rookie defenseman Devon Toews entered the lineup for the Islanders, upon Thomas Hickey’s injury. While Hickey (paired with Ryan Pulock, and typically used against top opposing forwards) was not playing poorly, the resulting pair of Toews and Scott Mayfield helped to balance out the group of six defensemen, immediately providing better play than the Adam Pelech / Mayfield third pair.

The purple dots below mark Toews’s entry. The first dot represents the last game the Isles played without Toews in the lineup. The second dot marks the conclusion of Toews’s first 10 games for Isles. (This second dot is important, since the blue and red lines are based on 10-game rolling average.... In other words, all the data between the two dots [on the x-axis] is from a mix of games without and with Toews in the lineup.)

Also credit Josh Ho-Sang (1.90 on-ice) and Michael Dal Colle (1.48 on-ice) for helping to keep the expected-goals rate down, during their games in the lineup.

Since Dec. 23, when Toews entered, the Isles have been first in the NHL for 5v5 expected goals-against per-60, at 1.81. The next best East team: the Montreal Canadiens, at 2.17. For actual goals-against Isles and Habs are first and second again: 1.43 and 1.90, respectively.

Here is another visual of before Toews (left side) and after Toews (right side). There was an immediate drop in expected goals-against per-60 at 5v5. (Keep in mind that average for NHL teams is 2.49.)

The Isles weren’t particularly great at keeping expected goals to a minimum before Toews arrived, though. The first 16 games we can see that while actual goals-against rate was 5th best in the NHL (2.00), expected goals were near-average (2.46).

In other words, the goalies were the difference at 5v5 for Islanders the first month, especially Thomas Greiss, with his .963 Fenwick save % (expected was .951).

Games 17 through 31 (November 13th through December 15th) the Isles hit a bit of a rough-patch defensively. They were 9th-worst for expected goals rate (2.75), while goaltending helped to keep the actual rate (2.45) 11th best in the league. This time it was largely due to Robin Lehner (.935 Fenwick save%, while .918 was expected), but Greiss was solid as well (.936, while .931 was expected).

The Isles also survived this run of games (scored 49% of 5v5 goals) largely due to their league-leading 7.7% expected Fenwick shooting. (DAL was next best, at 6.9%.) In other words, a unique percentage of their shots-for were high-percentage chances, as head coach Barry Trotz’s 5v5 system began to snap into place.

Games 32 through 43 largely overlaps with the introduction of Toews, but it also represents some of the best performances of Lehner (.969 actual, .951 expected) and Greiss (.952 actual, .932 expected).

The expected goals-against rate was a respectable 2.32 over this span, while actual was an incredible 1.53.

The next few games we can see that the Isles’ actual rate (red line) is actually more due to the expected rate (blue line, or a measurement of team-defense) than to outstanding goaltending. How do we know this? We can see that the blue line is closer to the red line than it is to the green line (NHL average).

The stats back this up: over the six-game stretch Isles were best in the NHL for expected GA/60 (1.56). But it is still good goaltending whenever the actual rate (red line) is lower than the expected (blue line).

These past few games we see that (while the expected rate may be leveling off— which at sub-2.00 would be phenomenal) the actual rate has some distance, meaning goaltending has been great again. Since the blue line is close to halfway between the green line and red line it is fair, I think, to attribute nearly half of the success to Isles as a team and half to goaltending, in terms of keeping the puck out of the net at 5v5.

Keep in mind this is 10-game rolling, so the visual of these past seven games (on the chart) also includes data from the previous nine games.

One of the big takeaways here is that since Toews stepped into the lineup (and Ho-Sang/Dal Colle) the Islanders as a team have been much better keeping expected-goals rate very low.

That said, goaltending has been remarkably consistent all season, as actual goals-against have been lower than expected (often much lower) over 10-game stretches each of the past 30 games.

While the goaltending doesn’t show any sign of faltering, even if they are near-average the rest of the season, the Isles could still keep 5v5 goals-against lower than average if they can maintain their stifling defensive play.

All stats used in this article were updated the morning of Wednesday, February 13.