Warning: small sample size ahead.
The purpose of making sports into a numbers game is not to sap the fun out of things, rather it is a quest for enlightenment and understanding.
There’s a credo from baseball analyst Mitchel Lichtman that goes something like, “I’m not sure, this is why I’m not sure, and this is roughly how not sure I am.” I find that does a good job at covering what advanced analytics do. They don’t exist to hate on your favorite player — unless it’s Dan Girardi — or prop up hipster-type picks like Mathieu Perreault. The numbers simply are what they are, and we can use them in conjunction with the eye test to learn more about the game.
I add that preamble because what is about to follow is not necessarily sympathetic towards the Vegas Golden Knights.
As of today, the Knights have scored nine goals and allowed just four. Based on the analytics, they are still looking pretty good, with an expectation of 13.05 goals scored and 8.98 allowed. More on how those numbers came into existence later.
What can we glean from that information? First, that Marc-Andre Fleury has been nothing short of incredible, and secondly that the Knights have not been scoring as much as they should.
While that 13.05 vs. 8.98 differential seems excellent — and it is, with the +4.07 mark ranking third in the NHL — there is a lot more at play. The results have muddled our view of the process, and a closer peek at the numbers reveals a team that is having a difficult time at 5-on-5.
The analytics community, myself included, often holds 5v5 numbers above all else for good reason: the bulk of every game is played during regulation without any players in the box and with both goalies in net. A team that excels at 5v5 and is poor with the man advantage will be better than a team with the attributes reversed.
Vegas, at least as far as expected goals is concerned, has the fourth-worst differential in the NHL at 5v5, registering a paltry 3.98 goals this season compared to allowing 6.16. On a per-60 minute basis, that’s just 2.03 goals versus 3.14 scored by their opponents. If you’re likely to give up one more goal than you score at even strength per game, well, you’re not going to win very often.
However, although outshooting a team 5-on-5 is indeed something to strive for, no two shots are created equal. That’s where “Expected Goals” comes into play. Expected Goals, or “xG,” puts an exact value on every shot, based on situation, handedness, type of shot and location. A 5-on-5 shot from the right point? Very low xG. A swift wrister from 10 feet? Much higher.
You see, although the Knights have been taking enough shots — Vegas has averaged 61.14 5v5 shot attempts per 60 versus a league average of 58.13 — their quality of shot has been poor. Per attempt, they are averaging one expected goal for every 30.15 shots taken at 5-on-5. This pales in comparison to the New Jersey Devils (?!?) at 15.41 and the Toronto Maple Leafs at 16.28. Basically, Vegas has to take nearly twice the amount of 5v5 shot attempts as New Jersey or Toronto to score the same amount of goals. However, Vegas is not the worst in the league—that dubious honor goes to the Detroit Red Wings, who need 32.65 5v5 attempts to notch an expected tally.
Hey, do you like shapes? How do you feel about hexagons?
The following is a hex-bin plot, more or less a heat map that shows success rate in relation to from where the puck was shot. As it would stand to reason, the closer to the net you are, the more likely the shot will result in a goal.
So that shape above is referred to as “home plate,” based on the shape of the baseball artifact that has similar dimensions. Home plate is the danger area for goalies and a shot from there is more likely than others to find twine. Let’s take a look at the first three games for Vegas:
Yikes. The Dallas Stars hammered them from in close at even strength. If it weren’t for the goaltending — Fleury being a monster and Ben Bishop getting hurt and replaced by a cold Kari Lehtonen — the game could have had a tremendously different outcome. While the Knights outscored Dallas 2-0 at even strength, the shot data shows it should have been roughly 2.73-1.07 in favor of Dallas.
This one was particularly interesting. Even though Vegas significantly outshot Arizona at even-strength — to the tune of 45-28 — their expected goal figure was just 1.65 compared to the Coyotes’ 1.62 mark. Again, the Coyotes took more shots from better areas and it just so happened the Flower was there to turn them away.
Finally, we have the most recent game, and while the Knights have done a better job at keeping Arizona out of the home plate area, they did not get many there themselves. Looking right around the crease, six of the seven most close-range shots came off Coyote sticks. Also, hey Erik Haula, whatcha doin’ all the way out there?
5v5 Shot Attempt and xG Data
And now, just for fun, here’s a look at which individual Knights have the best expected goals at 5-on-5 this season, as well as their overall xG mark:
Individual Golden Knights xG
|Player||5v5 ixGF||All Situations ixGF|
|Player||5v5 ixGF||All Situations ixGF|
Jonathan Marchessault, David Perron and William Karlsson have made it happen out there, but don’t have any goals to show for it yet. Trust the process, guys. James Neal appears to have gotten considerably lucky, but sometimes it’s better to be lucky than good. Sometimes.
Marchessault sits 97th in the NHL in 5v5 xG, while Neal and Perron are 23rd and 28th in all-situations xG. On the defensive side of things, Jason Garrison, Brayden McNabb and Luca Sbisa are in the bottom 26 defensemen (out of 145) in terms of all-situations xG (min. 50 minutes played). Oof. Nate Schmidt is third. Yay! More than half of his 1.26 expected goals came on his game-tying tally (0.667 xG) against Arizona with Fleury pulled. Does that mean that you or I would have scored there? Y’know...maybe!
Quantity is great, but so is quality. It’s the truly great teams though that do both. We hear “get pucks to the net” and “get pucks in deep” from players and coaches but there is a lot more nuance than that. Just like simply citing Corsi without context doesn’t paint a full picture either.
As far as the Golden Knights go, there is a lot to like. The team is drawing far more penalties than they take, and even though the goals have yet to start flowing in on the power play, they likely will. It’s only a matter of time, as the shot quality has been good enough to be worth 3.8 expected goals. They have also shown excellent work on the penalty kill and have benefited from stellar goaltending. Having said all that, if they do not pick up their 5-on-5 shot quality, it is likely to be a long season out in the desert.
All data from this article courtesy of corsica.hockey and hockeyviz.com. Both are invaluable for hockey fans and can be supported via their respective Patreons.