What we discovered from Connor Hellebuyck’s time in the NHL

With Ondrej Pavelec’s return to health, the Winnipeg Jets’ netminder prospect extraordinaire Connor Hellebuyck returned to the Manitoba Moose in the American Hockey League.

The self-proclaimed “big and boring” goaltender blocked shots for 26 games during his call-up to the NHL. Over that time he stopped 91.8 per cent of shots on goal and posted a 13-11-1 record.

What did we learn about Hellebuyck from his stay in the NHL?

Well, to be frank, there was really no new information to the goalie’s game that was not known before. For the most part, we just reaffirmed how legitimately good the young netminder is.

He still played his own game: big and boring. He was a calming factor in the net, never relying on athleticism. He played big in the net, cutting down the angles and stopped pucks. In many ways it was a stark dichotomy from the style of goaltending familiar to Jets’ fans.

Those who followed Hellebuyck since the Jets’ selected him 130th overall in the 2012 NHL Entry Draft always recognized that Hellebuyck had a shot at being exceptional. His history, after all, has been exceptional.

He was drafted out of the NAHL where he posted an impressive 0.930 save percentage, and took home multiple awards such as NAHL Goaltender of the Year and Rookie of the Year. He spent the next two seasons in the NCAA playing for the UMass-Lowell Riverhawks, sending his team twice to the national championships tournament. He carried a 0.946 NCAA save percentage (a NCAA record), a 38-12-2 record, and a total of 12 shutouts over his collegiate career. He won more awards than we care to list, including the inaugural Mike Richter Award for best collegiate goalie. He has since played 70 games in the AHL, while stopping 92.3 per cent of 2209 shots faced.

But that is only what he accomplished outside of the NHL.

In his short time in the NHL, the 22-year-old posted a 0.918 save percentage, stopping more than the average number of shots a NHL netminder or starter tends to stop. We could finish our discussion there, but we tend to like a more in-depth research into player performance at Jets Nation.


The top table shows Hellebuyck’s performance over his Jet call-up. Numbers are meaningless without context, so we’ve added the most familiar and commonly used goaltender on the Jets: Ondrej Pavelec.

Save percentage is simply the percentage of shots on goal that does not enter the net. This number is a much better measure of performance than GAA or wins, as the goaltender input in those two numbers are in fact save percentage. A goalie impacts wins by keeping goals against down as much as possible, and GAA is simply shots against per 60 minutes multiplied by 1-sv%.

Save percentage is the ultimate goal for a netminder, much like goal differential would be for a skater. As we noted earlier, Hellebuyck has stopped a greater percentage of shots on goal than one would expect the average NHL netminder or average starter to pull off.

We can further split Hellebuyck’s performance depending on shot location: low danger (perimeter), medium danger (high slot), and high danger (low slot). We see a legitimate trend in performance between the three areas. This fact should help us set our expectations depending on where the shots are coming from, which is depending on the team’s effectiveness in keeping shots to the outside. This is a major portion of the shot quality bullet people are looking for with goaltending and defence.

However, WOI’s Adjusted Save percentage, a goaltender’s save percentage adjusted for shot location differences, shows us that the differences experienced from league normal distributions is fairly small. Both goaltenders receive a boost, but minimally so.

Hellebuyck outperforms Pavelec in all areas except in stopping perimeter shots, and the gap in overall performance is about that of a slightly above-average starter versus a backup netminder.

The comparison may not be fair though, as Hellebuyck has played with the Jets exclusively this season, where the Jets have been heavily penalized relative to their early years. The lower table contains the same numbers exclusively at even strength, which reduces sample but evens the playing field.

Nick Mercadante, of Hockey-Graphs fame, takes a similar look to WOI’s AdSv% with his own Adjusted Goals Saved Above Average per Sixty statistic:


Here we see that Hellebuyck’s performance has been far above average, while Pavelec and Hutchinson have been similarly equal in their 5v5 performance and below average.

(Note: Hutchinson’s performance at 5v5 is below Pavelec’s, despite out performing overall due to superior penalty kill performance. IE: Pavelec has been better at 5v5 while Hutchinson has been better at PK and over All Situations)

DTM About Heart and Asmae, also of Hockey-Graphs, have also developed a way to look at goaltending through a fairly sophisticated shot quality model:


DTMAH and Asmae use their expected goal model to predict what a goaltender’s 5v5 Fenwick Save Percentage should be (Fenwick means missed shots are included, not just shots on goal). We then compare goalies by how many percentage points their save percentage is greater than the expected (xSv).

Hellebuyck has already stopped about 7 more goals (xGS) than we’d expect the average netminder, meaning that Hellebuyck has already provided a greater positive impact than Pavelec has over his career.

Now, there is a common (and sometimes misunderstood) saying that goaltenders are voodoo. This does not mean that goalies are random. The saying came about due to how volatile goaltender performance can be and how much longer it takes performance to regress to true-talent level on average. This regression can sometimes confuse fans, media, and hockey personnel, sometimes causing poor evaluation of talent, over payment on performance, or false belief of collapse or new emergence.

Small sample, however, does not mean that performance is meaningless in the short run; it just means that the interval of confidence is much larger. Nick Emptage, again from Hockey-Graphs, displayed this comparing what the 5v5 save percentage we’d expect 95 per cent of goalies with a true talent of either 0.925, 0.915, or 0.905 to post given the sample size:


Another way to look at this is the better a goalie performs in that short run, the more likely he is a good goaltender. Or, a bad goaltender can post good numbers in the short run, but is more likely to post bad numbers than good, with the opposite being true for a good goaltender. When you get to a larger sample though, you begin to have a pretty good idea of where their talent falls.

An example, a team with a goaltender who posts a 0.925 save percentage over 400 shots has about a 61 per cent shot he’s legitimately a 925 talent or better, then 24 per cent chance he’s about a 910 talent, and then the rest being a 905 talent or worse.

Hellebuyck meanwhile has stopped 0.939 of 524 shots.

So what do we know about Hellebuyck?

Well, he’s been exceptional at every level. His 0.918 save percentage in the NHL is the first time he’s a sub-920 save percentage since being drafted. He has performed above expectations both for the average netminder and given the shots he has faced. He may not always continue to be as good as he has been, but he’s more likely a legitimate starter or better than not. He’s also a full three years younger than the average peak age for goaltender performance.

In other words, we’ve only found one more thing to add to the growing list of reasons why Hellebuyck is likely a very, very good goaltender.

All numbers courtesy of WAR-on-ice.com unless otherwise stated.