Over the course of the 2019/20 Premier League season, there has been a certain level of criticism aimed at David de Gea, and many questions have been raised as to whether it is time to let Dean Henderson have a run in the number 1 spot when he returns from Sheffield United. At the Blades, he has impressed massively and has seemingly become a leading goalkeeper in the league. Does he deserve to have a shot at being the number one stopper at United? Or is de Gea being undeservedly scapegoated for his performances?
Over the course of this season, David de Gea has played and started in all 29 of Manchester United’s Premier League outings. In those 29 games, he has conceded 30 goals, which of all the goalkeepers to have played at least 18 matches places him 7th out of 18 keepers. Dean Henderson is 2nd with 22 goals conceded but has played 27 matches instead. Then when it comes to save percentage, de Gea’s is 71.3% whereas Henderson’s is 75.8%. On the face of this, it would seem that Henderson has definitely been the better goalkeeper. However there is a glaring issue with this statistic. It doesn’t take into account the quality of shots being faced. So where de Gea is saving shots at a slightly lower rate, he could be facing much more difficult shots. Thankfully, there is a statistic that can try and help us remove some of this uncertainty; Post-Shot Expected Goals minus Goals Allowed. Working in a similar vein to how Expected Goals can be used but using only shots on target as these are the only shots the keeper interacts with, by using post-shot expected goals, we can see how many goals the goalkeeper should have conceded based on the quality of shots the player has faced. For the rest of this article, I will refer to Post-Shot Expected Goals as ‘PSxG‘.
Over the course of the entire season, de Gea’s PSxG was 28.9 and Henderson’s was 28.4, so de Gea would have been expected, on average, to concede 0.5 more goals over the entire season, so barely any difference in that regard. Another way in which we can look at this by adjusting PSxG per shot on target. This then adjusts for the number of shots on target each keeper faces. For Henderson, his PSxG per shot on target is 0.32 and de Gea’s is 0.29, so there isn’t much difference at all in the quality of each shot being faced. Remembering that David de Gea has played two more games, adjusted to per 90 metrics, de Gea has faced shots that should equate to 1 (0.9966 to 4 d.p.) goal per 90. Henderson on the other hand, should’ve conceded 1.05 goals per 90. Not a massive difference between them at all, but Henderson has conceded 8 less goals than de Gea, which suggests that Henderson has performed much better than the Spaniard, even when taking into account quality of shots faced. When we look at the number of goals conceded compared to expected, Dean outperforms by 6.4 goals whereas David is -0.1. This means Dean has saved his side just over 6 goals this season, where de Gea has performed completely in line with how the model would expect the average to be.
We will now look at aspects other than shot-stopping, as today more so than ever, being a goalkeeper is much more than being able to save shots well. Firstly when looking at their passing stats, we have to take into account the vast difference in how each keeper is instructed to distribute the ball. Where at Manchester United, the keeper looks to often play short and keep possession, at Sheffield the keeper will look to play long, whether from open play or goal kicks. With this in mind let’s take a closer look. Of David de Gea’s attempted passes, 35.6% of his passes were longer than 40 yards; in comparison Dean’s percentage was 78.0% of his total passes! Of these launched passes though, David actually completes them at a higher rate of 36.9% to Dean’s 28.0%, some of this will be down to how each team plays and the opposition, but nonetheless, de Gea completes them at a higher rate. However, because Manchester United tend to attempt these style of passes much less (which can be seen by our average goal kick distance being 36.4 yards to Sheffield United’s 76.1 yard average goal kick) it is more pertinent to look at the closer range passing. For passes between 5 and 25 yards in distance, David completes them 99.1% of the time and Dean 98.7% of the time, so no significant difference whatsoever. So whichever keeper plays in goal next season, we would likely not lose or gain much in terms of passing. When looking at command of area with crosses, both players have faced a similar number of crosses (David – 222, Dean – 203) but Dean has successfully stopped 7.9% of the crosses into the penalty area, compared to David’s 3.6%. As a comparison to these two players, Liverpool goalkeeper Allison stopped crosses this season 7.0% of the time. This suggests that Dean is commanding in the box and has no fear contesting crosses. Also this season, de Gea has made 2 errors leading to shots compared to Dean with 0. Finally, we look at the number of defensive actions each keeper makes outside of their penalty area as the sweeper keeper role becomes more and more prevalent in the sport. Per 90 minutes David has completed 0.31 defensive actions outside the are compared to Dean’s 0.48 (for context Allison has made 1.19 per 90). The United loanee completes slightly more per 90, but both men don’t come close to the Liverpool stopper.
The shot stopping stats suggest that Dean is the clear choice to be our number 1 from next season. However, football is never as simple as just statistics. We have to remember David is a keeper, who only a few seasons ago was the best in the world, and in 2017-18 had one of the best seasons as a goalkeeper we have ever seen, both statistically and using the eye test. The question which is more important is has de Gea not done enough to deserve more time and patience from the club? Some people would argue we have to act quick before the rot develops even more, I’m not so sure he’s finished just yet.