Ziggdiezan
Registered User
- Apr 10, 2015
- 10,847
- 5,676
Expected goals against does a poor job of accurately representing things like odd man rushes which the leafs give up a shit ton of.So the Leafs are middle of the pack in expected goals against while Keefe has been the coach yet they are 24th in goals against per game during this time. Since xGA takes into account the percentage of likelihood of a goal depending on shot location, shot type, etc., then yeah, the goalie has actually been a pretty big factor for worse.
For example, on a 2 on 1 whereby one player plays a cross seam pass and the other players takes a shot near the haskmarks. Assuming the same player (shot quality is used in xG) takes a shot from a similar location but say during a cycle whereby there is a tons of coverage, both of these shots would be given a very similar expected goals even though as we know the 2 on 1 is much much more likely to resulted in a goal.
Just one of the faults with expected goals along with the fact the data sets they used to perform all their regressions was before a lot of the recent changes (goalie equipment etc) and how they handle shot quality isnt my favorite. Inflates the expected goals of historically good shooters creating a self-perpetuating situation. For example if Auston Matthews and Mitch Marner shoot from the exact same spot, Matthews will automatically have a higher expected goals. Makes it more difficult for newer/playmaking players to have a comparably good expected goals contribution