I read a tweet a few weeks ago from a Liverpool fan about how opposing goalkeepers always seem to play their best match against Liverpool, and citing the fact that five opposing goalkeepers had been selected Man of the Match this season. StatDNA has a statistic that measures how many goals a goalkeeper was expected to let in, given the quality of the finishes he’s faced, so I decided to use this statistic to see how valid the claim was. In mathematical terms we’re measuring the expected value (EV) of the shots on target that the goalkeeper faced (excluding penalties and own goals). By subtracting actual goals from expected goals, we can get a measure of how well a goalkeeper performed in a particular match. Note that this statistic only measures a goalkeepers performance against shots on target, not other aspects that are clearly important like the ability to claim crosses, distribution, etc.
The chart below shows that Liverpool may indeed have a gripe. There were 19 matches in which a goalkeeper displayed outstanding performance by reducing the opposition’s expected goal total by 1.5 goals or more, and five of them came against Liverpool (through round 19 matches). Liverpool’s not the only team that’s been damaged by the opposing goalkeeper, Manchester United clocks in with 4 occasions and then a host of other teams have 2 and 1 occasions. Typically, higher quality teams are going to suffer this fate more because they get more shots on target, thus giving the opposing goalkeeper more chances to be a hero. It is interesting to note that neither Manchester City nor Chelsea have suffered a single outstanding goalkeeper performance by our measure.
If we add up Liverpool’s suffering – the goals that they would have scored in theory against average goalkeeper performance – it’s more than 8 goals additional. The next two most affected teams are Fulham and Everton, each clocking in at around 5 goals. Manchester United has also had some poor goalkeeper performances against them so on overall net goals due to opposing goalkeeper performance; they are below Everton and Fulham.
One could reasonably expect a higher-scoring second half of the season for the most-affected teams if the opposing goalie performance truly is random (which one would think to be true) and if the team performance remains similar (which obviously is more questionable, given injuries, suspensions, etc). Keep in mind these are rough statistical measures (which we also improve over time as we collect more data), so to anticipate 8 more goals in the second half for Liverpool, ceteris paribus, would certainly be a leap
.
Now, while Liverpool certainly may have experienced some bad luck, their low goal total is actually *mostly* of their own doing. While Manchester United have also suffered from some cracking performances by opposition goalkeepers, they at the same time lead the league in their finishing quality and are in the top-4 in shot quality. Liverpool, meanwhile, rank in the bottom half in both finishing and shot quality (to read more about how we measure finishing and shot quality, click here and here). Differences in finishing quality and shot quality actually explain 67% of the 25 additional goals that Manchester United has scored over Liverpool in the Premier League this year.
The last point is that Liverpool’s point total in these excellent goalkeeper performance matches was 1.8 per match, actually slightly better than their 1.7 average in 20 EPL matches. The same goes for Manchester United who earned 2.5 per match against top goalkeeper performances vs. 2.25 in their first 20 EPL matches. These totals are unadjusted for the quality of opponent, but at a high-level, point totals may not have been greatly affected.