Overrated Stats: Clean Sheets

At the end of each season, it is customary to award the goalkeeper with the most clean sheets the Golden Glove or some version of that award. This is usually used to distinguish this keeper as the best in the competition. However, there is a problem with using cleans sheets to measure a goalkeeper’s competence and this lies in the fact that the stat fails to take into account all the other factors that go into achieving a clean sheet in any given game. The defenders always have a huge part to play in preventing goals due to all the clearances, tackles and blocks they make so it is entirely possible for a keeper to go a whole game without having to save a single shot. So, how exactly are we meant to measure how good a goalkeeper is if not with clean sheets?

The answer to this question could lie in the use of a statistic called Post-Shot Expected Goals (PSxG). xG provides a rating for each shot based on the historical precedent of the probability of shots from any given location ending up in the back of the net. However, xG doesn't take into account what happens after a shot is taken and this is where Post-Shot Expected Goals comes in. What PSxG does is combine the shot quality with the location a shot is supposed to end up in to assign it a value. For example, a shot from outside the box destined for the top corner may have an xG value of 0.02, but the PSxG will probably be closer to 0.60 since the shot is heading for the top corner which is harder for a goalkeeper to reach. PSxG ignores stats like save percentages and clean sheets and only seeks to quantify a goalkeeper’s shot stopping ability. If the goalkeeper is facing shots with higher PSxG values, then he is facing shots that are more difficult to save.

A useful graph for compare keepers from the Premier League this season would be a scatter plot comparing a keeper's PSxG minus goals allowed per 90 minutes to his total goals conceded. A positive value means you are saving more shots than you are expected to and vice versa for a negative value. With this, we could measure a keeper's shot stopping ability based on what they have shown so far this season. We could also adjust the size of each point to indicate how many shots on target they have faced so far in the season in order to add much needed context to their performances.

 [Data from FBref]

   

We can see who the top performing goalkeepers in the Premier League are so far this season based on this graphic. Emiliano Martinez of Aston Villa seems to have solidified himself as one of the best keepers in the league so far as he not only saves more shots than he is expected to per game (>0.35 PSxG/90), he is amongst the leaders in least goals conceded. Nick Pope is also making a strong case to be England's new number one as the numbers show his performances don't only look impressive on the TV, they are also impressive on the data sheet (>0.3 PSxG/90, 13 goals conceded). On the other side of the spectrum, we have the likes of Aaron Ramsdale, Vicente Guaita and Rui Patricio who are all conceding more goals than they are expected to and are conceding the most goals in the league. 

The beauty of this graph lies in the fact that you can identify some keepers' performances that may have gone under the radar if you don't factor in all three categories. Alphonse Areola, Sam Johnstone, Illan Meslier, and Karl Darlow all deserve a mention as even though they are amongst the keepers that have conceded the most goals in the league, they have also faced the some of the most shots in the league. Despite that, they are still conceding less goals than expected per 90 with Areola being the most impressive having saved more than 0.3 more goals than expected per 90.

It is also interesting to note that this graph shines more of a light on goalkeeper's that you would not expect to be in line for the Golden Glove. Alisson, Hugo Lloris and David de Gea are usually mentioned amongst the best of the best purely based on the clean sheets they are able to accrue over the course of a season. These keepers may concede less since they play behind quality defensive lines, so the true test for them would be to look at their PSxG per 90 and from that, we can see that while Alisson, Lloris seem to be saving more than expected, de Gea is letting in slightly more goals than would anticipate a goalkeeper of his calibre to do. Based on what you are looking for, there are a plethora of conclusions you can draw from a graphic such as this.

Post-Shot Expected Goals is an example of statistics done right. Unlike clean sheets where goalkeepers take all the glory that is meant to be shared between themselves and their respective defensive lines, PSxG isolates the keepers and sidelines every other arbitrary measure that may get in the way in order to measure a goalkeeper's competence as it pertains to shot stopping. With the growing popularity of data in sports, it's only a matter of time before clean sheets are replaced by stats such as PSxG in the eyes of the general public.