What are the Luckiest and Unluckiest Teams in the NHL Right Now?
After my recent journey to build a better model to estimate an NHL team’s winning percentage came up short, I was stumped. Hockey analytics has lagged behind the analytic booms in other sports partially because it is more difficult to actually analyze from an unbiased perspective, and failing to build a better model to predict a team’s winning percentage using a combination of basic and advanced statistics is representative of that (albeit in a minor way). I think that makes it more exciting.
Vox has a great — while being very much subjective — graphic demonstrating the difficulty in using analytics due to how NHL games are influenced by luck more often than other team sports like baseball and basketball:
This got me curious about measures of luck in hockey analytics, so that’s exactly what we are going to focus on for this article. The main measures we are going to be using are PDO and Goals Differential Above Expected. PDO is not an acronym for anything in particular, but it represents the following formula:
PDO = (GoalsFor/ShotsFor)+(GoalsAgainst/ShotsAgainst)
A score of 1 means the team is performing exactly as they should. These teams are neither lucky nor unlucky. Teams under 1 are considered unlucky, where teams over 1 are luckier.
That leads us to the next measure. Goals Differential Above Expected is calculated by the following formula:
GDAE = (Goals For Above Expected)/(Goals Against Above Expected)
A score of 0 means the team is performing exactly as they should. Teams above 0 signifies teams that are more lucky, whereas teams under 0 are more unlucky.
For this article, I am using PDO data from Natural Stat Trick and Goals Differential Above Expected data from MoneyPuck.com. Keep in mind that this data is also from before the games scheduled for April 19, 2021, so it is subject to change.
- Washington Capitals — PDO 1.022 (2nd), GDAE: 20.48 (2nd)
- Vegas Golden Knights — PDO: 1.020 (3rd), GDAE: 16.36 (4th)
- Minnesota Wild — PDO: 1.026 (1st), GDAE: 11.26 (7th)
- New York Rangers — PDO: 1.018 (5th), GDAE: 16.88 (3rd)
- Winnipeg Jets — PDO: 1.012 (9th), GDAE: 24.46 (1st)
- Philadelphia Flyers — PDO: 0.976 (2nd), GDAE: -23.58 (2nd)
- Ottawa Senators — PDO: 0.975 (1st), GDAE: -21.48(3rd)
- San Jose Sharks — PDO: 0.980 (3rd), GDAE: -23.96 (1st)
- Buffalo Sabres — PDO: 0.984 (4th), GDAE: -19.26 (4th)
- Columbus Blue Jackets — PDO: 0.984 (5th), GDAE: -13.2 (7th)
It is easy to see Vox’s graphic panning out according to these rankings — luckier teams have tended to see better results in the standings compared to the unluckier teams, who are near the bottom of their respective divisions. That’s not to say the standings are completely indicative of luck, but there is a noticeable correlation that shows luck certainly has some aspect of a team’s winning percentage.
Conventional thinking would point lucky teams getting less lucky as the season goes on and the unlucky teams would start to get less unlucky. In effect, it’s also saying that the lucky teams are not as good as they might seem in the standings and the unlucky teams are not as bad as they might seem. Is there weight to this thinking?
Last season, the luckiest teams were the Colorado Avalanche, St. Louis Blues, New York Rangers, and Boston Bruins. The Rangers lost in the play-in round (and ended up winning the draft lottery and selecting Alexis Lafrenière); the Blues lost in the first round in an upset to the Vancouver Canucks; the Bruins lost in the second round to the eventual Stanley Cup Champions, Tampa Bay Lightning; and the Avalanche lost in the second round in an upset to the Dallas Stars. Two of these were considered upsets, but it stands to reason that their luck just ran out. Luck might have a bigger place in hockey than other major team sports, but that doesn’t mean it is an aspect a team can count on every night like they can count on Auston Matthews to put one in the net.
It’ll certainly be worth revisiting this piece along with my earlier article about predicting teams and their winning percentages as of last month to see which way is better in predicting potential upsets in the playoffs.
Let me know your thoughts, or any suggestions for topics to analyze in the upcoming weeks!