Assessing Goalkeepers and Goal Scorers the Liverpool Way
Scorers should be recognised for beating the best goalkeepers, with shot stoppers acclaimed for keeping out the top finishers. Who's the best in the Premier League?
Premier League goalkeepers are rubbish. The data says as much.
Between 2017/18 and 2023/24 they collectively conceded 7289 times (excluding own goals) from a total of 6976.4 expected goals on target (xGOT). That’s a league-wide underperformance of roughly 45 goals a season.
It could be that it’s the data that’s to blame rather than the shot stoppers. Adding France, Germany, Italy and Spain to the sample shows that goalkeepers in the big five leagues underperformed by 941 goals across a seven year period. Can’t anyone save a shot around here?
Perhaps the model is wrong. Or maybe it’s a quirk of an enormous sample; the average goalkeeper being just 0.04 goals per match below par makes the system sound more accurate than underachieving by almost 1,000 goals.
The heart of the issue is that it is impossible to beat the model year after year. This became apparent when reviewing the efforts of Alisson Becker and Giorgi Mamardashvili in 2025/26. Both are underperforming against xGOT this season when they have rarely had such problems in the past:
There are 68 goalkeepers who featured for over 900 minutes in at least five campaigns in the aforementioned seven year period. Not one of them saved more goals than expected every season. You need a huge sample to get a decent read on their abilities.
If anyone has the data for 2024/25, let me know. For now, I’ve pieced together 2017-2024 with the current campaign. The stats appear to pass the eye test for the big dogs of the goalkeeping world; Jan Oblak has performed at 14.8 per cent above xGOT, with Thibaut Courtois is at 10.5 and Allison 8.6. The latter would look better with last season included but perhaps the others would too.
The Brazilian’s statistics brought to mind a passage in Ian Graham’s book, How To Win The Premier League. When he was Director of Research at Liverpool, they would use similar numbers to weight performance.
“Each outfield player and each goalkeeper was given a conversion skill. Players that consistently overperformed the predictions of our Post-Strike model increased their skill rating and vice versa. The result was that elite finishers like Son Heung-min and elite shot-stoppers like Alisson Becker were expected to score and save more goals than the model estimated. A side effect was that making a save from a Son shot, or scoring past Alisson, was more highly rewarded than performing the same action against a lower rated player.”
By compiling the details of every goal scored in the 2025/26 Premier League, we can do a simplified version of the same process. Who deserves credit for against whom they’ve scored and which goalkeepers might have exceeded expectation given who has beaten them?
