The expected goal is a darling statistic of the analytics movement, but what exactly is it? And what can it tell us about the Toronto Maple Leafs?
Expected goals can explain a lot about the Toronto Maple Leafs.
Expected goals (xG) are a measurement of two things: shot quantity and shot quality. Shot quantity is simply gathered as shot attempts, but determining shot quality is more complicated.
To measure shot quality, a compilation of NHL data looks at all of the different types of shots that are taken at varying distances and angles in relation to the net and calculates the average odds of an NHL player scoring in any given situation.
For example, according to moneypuck.com, a slapshot from 60.2 feet at a 11.5-degree angle during 5-on-5 play goes in an average of 2.1% of the time, resulting in 0.021 xG. A higher quality shot, such as a backhand shot from 7.6 feet and 23.3-degree angle during 5-on-5 play, goes in an average of 18.0% of the time, resulting in 0.18 xG.
Team Expected Goals
The xG from all shot attempts can be added over the course of games or seasons. In the context of team stats, there are two important statistics:
xGF (expected goals for) – A measure of shot quality and quantity for a team
xGA (expected goals against) – A measure of shot quality and quantity against a team
This season, the Toronto Maple Leafs accrued 141.98 xGF (ranked 3rd) and 133.48 xGA (ranked 21st), indicating that they outmatched their opponents in producing a high quantity and quality of scoring chances.
xGF does not always align with a team’s GF because in real life, players may have more shooting talent than other players, and luck is a factor. Expected goals, however, will still better predict the future than real goals will.
For instance, the Tampa Bay Lightning and Dallas Stars sit relatively close in xGF in the 2019-20 season, with the Bolts ranked 9th (131.77) and the Stars ranked 13th (128.72). However, the sharpshooting Lightning rank 1st in GF (162) while the Stars rank 30th (108).
The difference in reality comes from Tampa leading the NHL in team shooting percentage, while Dallas is 30th. Tampa almost certainly has more shooting talent than Dallas, but over many simulations such a large difference would likely prove to be anomalous.
You can’t assume that expected goals will predict the future accurately, as that isn’t the point.
They are better utilized as a way to understand a team’s scoring chance creation, scoring talent, defensive play and goaltending. This will be covered in much further detail in a future article in this series.
Individual Expected Goals
To measure the performance of individuals, we can utilize the following two statistics:
ixG (individual expected goals for) – A measure of shot quality and quantity for an individual player
on-ice xGA (expected goals against) – A measure of shot quality and quantity against while a player is on the ice
ixG is most typically expressed as a rate over 60 minutes at 5-on-5 to correct for variations in ice time, games played, or situational player use (power-plays and penalty kills).
ixG/60 at 5-on-5 is the gold standard of expected goals statistics. This stat can be used to compare a player’s ability to generate a high quantity and quality of scoring chances to previous years, or to other players.
For example, Kasperi Kapanen’s ixG/60 dropped from 0.81 in 2018-19 to 0.51 in 2019-20, suggesting that he has struggled with getting scoring chances. This helps explain why his goal totals with the Toronto Maple Leafs have decreased this season.
As the gold standard, ixG/60 has a vast range of other applications that will be explored in further depth in a future article. Spoiler alert: It makes Auston Matthews look pretty good.
On-ice xGA (typically expressed as a rate) is a less poignant statistic than ixG/60. As with corsi and other on-ice statistics, on-ice xGA/60 can’t isolate for the impact of a single player and is dependent on the other nine skaters on the ice.
However, on-ice xGA/60 can illuminate trends when viewed in conjunction with other on-ice statistics and is most appropriate to use for assessing defensemen.
Expected goals are an integral part of performance assessment, particularly xGF and xGA for teams, and ixG/60 at 5-on-5 and on-ice xGA/60 for individuals.
As with any advanced statistic, context and interpretation of expected goals statistics are paramount to understanding what they can tell us about teams and players.