In hockey, expected goals (xG) are an advanced counting metric that uses past information of previous shots taken and gives context to derive shot quality of new shots. Using various inputs from previous shots such as shot location, events leading up to a shot, and league shooting averages spanning many seasons, every new shot taken can be assigned an expected goal value, regardless of whether the shot turned into a goal.
Different xG models created by the hockey analytics community take various metrics into account, but most publicly available models all aim to achieve the same thing of determining shot quality. As the number of variables/inputs accounted for each model will differ, so too will the xG values. When using xG to compare teams or players, it’s important to use numbers from the same model and to state which model was used!
In short, xG provides valuable information to better the understanding of shot quality, and helps advance the understanding of the sport overall.
Expected goals vs goals performance banners
To compare how players have fared with respect to the differences in their expected goals versus their actual goal outputs, I created a data visualisation using parallel coordinate charts or slope charts to get a sense of which teams have are enjoying immense goal scoring boosts with their overperformers and which teams are snake-bitten with underperformers.
The parallel coordinates provided plots every single player’s individual expected goals (ixG) to their actual goals and are connected by a line. The left hand columns show ixG totals, and the right hand columns plot total goals. The slope of the line indicates whether a player is scoring more or less than expected.
The more positive the slope (i.e. up and to the right), the more they’re overperforming. The more negative the slope (i.e. down and to the right), the more they’re underperforming. If a line is flat, then a player is scoring right about where they are expected to.
For the visualisation, xG data was obtained from NaturalStatTrick.com, and contains data from all games up to and including those ending on February 26, 2021. The plots were created in R using “ggplot2” and “GGally” packages, and edited on Adobe Illustrator. Click on the image to see the full size.
Insights
Better than Expected
Here are the top ten players with the most goals scored above expected.
Player | Team | Goals | ixG | Goals scored above expected |
---|---|---|---|---|
Auston Matthews | TOR | 18 | 8.73 | 9.27 |
Tyler Toffoli | MTL | 12 | 5.7 | 6.30 |
Dustin Brown | LAK | 11 | 5.36 | 5.64 |
Logan Couture | SJS | 10 | 4.95 | 5.05 |
David Pastrnak | BOS | 9 | 3.96 | 5.04 |
Patrik Laine | CBJ | 6 | 1.00 | 5.00 |
Joe Pavelski | DAL | 10 | 5.08 | 4.92 |
Nikolaj Ehlers | WPG | 10 | 5.10 | 4.90 |
Jonathan Huberdeau | FLA | 7 | 2.16 | 4.84 |
Josh Anderson | MTL | 9 | 4.21 | 4.79 |
Worse than expected
Similarly, here are the bottom ten players.
Player | Team | Goals | ixG | Goals scored above expected |
---|---|---|---|---|
Rickard Rakell | ANA | 1 | 6.56 | -5.56 |
Taylor Hall | BUF | 1 | 5.47 | -4.47 |
Mika Zibanejad | NYR | 2 | 6.30 | -4.30 |
Jeff Skinner | BUF | 0 | 3.43 | -3.43 |
Brady Tkachuk | OTT | 8 | 11.40 | -3.40 |
Viktor Arvidsson | NSH | 2 | 5.17 | -3.17 |
Adam Gaudette | VAN | 2 | 4.95 | -2.95 |
Derek Stepan | OTT | 1 | 3.82 | -2.82 |
David Kampf | CHI | 0 | 2.78 | -2.78 |
Sam Bennett | CGY | 3 | 5.55 | -2.55 |
Team observations
There are some observations that can be made about teams and divisions as well:
- The North Division is a goal scoring haven, with most teams having at least a few players heavily outscoring expectations aside from the Ottawa Senators.
- The Calgary Flames and Colorado Avalanche have a handful of players that have yet to score despite getting decent xG values with some players exceeding 2 xG to 0 goals.
- The Philadelphia Flyers and Washington Capitals are two teams that have quite a number of positive slope players with few negative slopes, likely meaning the teams are overall exceeding goal scoring expectations.
- The Detroit Red Wings, Anaheim Ducks, Buffalo Sabres, and Nashville Predators barely have any offence, with many players posting both low xG and low goal totals.
There are plenty more observations that can be made of course, these are just a few that stood out.
Expect the unexpected
It goes without saying that in a season as unique as this, literally anything could happen. The team realignments were bound to set off some teams in terms of lighting up the goal light game in and game out. Some teams were due to face easier competition than ever before, and some received the short stick instead.
As the season progresses, looking at expected goals can continue to shine a light on over and underperformers, and there will be lots to be learned from this shortened hockey season, both from team and fan perspectives.
You must be logged in to post a comment.