smarterscout SPECIAL REPORT: Shooting and saving skill
Measuring shooting and saving skill is one of the most challenging – and important – tasks for football analytics. For several years, analysts have tried to adjust their ratings for the difficulty of the shots being taken. But as far as we know, our platform is the only one that also adjusts for the skill levels of strikers and goalkeepers, in our case by rating them simultaneously. And this has generated some fascinating insights.
We rate players based on their chances of scoring or conceding from baskets of shots where the chance of at least one goal being scored is 50%. (You can read more about our method in the FAQ by clicking the link at the top of this page.) All players start out at par in our ratings, and we publish the ratings when we have enough data to gauge their skills. Let's take a look at the numbers behind the metrics:
Shooting and saving skill are not immutable. Both strikers and goalkeepers can have hot and cold streaks. Here is how our assessment of Alexandre Lacazette's finishing skill for non-headers in open play – the biggest category of shots – has evolved since he came to the Premier League:
Our system kept adjusting Lacazette's rating downward through most of his first season at Arsenal, but things turned around for him late in the 2017-18 season. Was it the arthroscopy that cleaned out his knee or something else? Regardless, his rating continued to rise until it peaked at around the same time in the next season. Since then, it's been dropping again, despite no notable changes in his status. There might be a kind of "hot hand" effect at work here; maybe he's just a streaky player.
The same seems to be true for Lacazette's teammate Bernd Leno. Here's how Leno's evaluation for stopping the same category of shots has changed since he came to the Gunners:
Leno's rating kept climbing throughout his first season. In other words, our system kept finding that his performance was better than his rating might have suggested. But in his second season in England, he wasn't quite as successful. The system adjusted his rating downward until he leveled out at a lower but still respectable level.
The dynamics of shooting and saving skill are not the same for every player. Unlike Lacazette and Leno, some players are consistently strong or weak. Look how Eden Hazard's rating for finishing skill kept increasing until it plateaued:
Hazard was a consistently good finisher in the Premier League, at least during the period we covered. In our system, his ratings rose until the results of his shots – handicapped using his rating – were no longer surprising anymore. At that point, his ratings neither rose nor fell very much; the system had a bead on him.
By contrast, Jamie Vardy seems to keep improving every season, so much that his finishing skill has yet to plateau through all our data:
Tactics matter. Sergio Aguero has never been the best finisher in the Premier League. He's not bad, either, but his talent is more about creating shots – and he usually finishes them at par. He does have hot spots on the pitch, though, and the expression of his finishing skill has been influenced by how Pep Guardiola has used him for Manchester City.
First, let's have a look at Aguero's shot charts for non-headers in open play during 2017-18 and 2018-19. Note how his shot volume on his right side changed:
Aguero does have unusual skill in finishing shots from wide positions on his right side, but in 2018-19 he had fewer opportunities from that area. In fact, the bulk of his shots were shifted to his left, and that may have resulted in less natural positioning for his striking. And in our data, his overall finishing took a big dip during that season:
Aguero's shot quality – as estimated for a generic striker – actually increased in 2018-19 when his shooting locus moved more towards the middle of the goal, but the percent of shots he scored fell. He's not a generic striker, and positioning matters.
Psychological effects are real. Coaches and scouts say it all the time: finishing and shotstopping require confidence. The data suggest that they are correct. Indeed, the effects of a player's emotional state on actions involving shots, where so much is at stake, can be downright startling. Here is how Kepa Arrizabalaga's shotstopping rating evolved after he came to Chelsea:
Kepa started out just fine, stopping non-headers in open play at an average rate or better. And then, about two thirds of the way through the 2018-19 season, something happened – something that sent him into a downward spiral from which he still has not recovered.
As Chelsea fans will no doubt recall, Kepa refused to be substituted by Maurizio Sarri in the Carabao Cup final, and then was dropped for the next league game as punishment. The psychological impact on the young GK appears to have been immense. Have Chelsea tried to help him recover, or will they simply give up on him by bringing in a replacement this summer?
The psychological aspect may hint at why the streaks we illustrated above aren't as common in our ratings for aerial and ground duels. These are more regular actions on the pitch. Because it's rare for them to lead directly to goals, they may not loom as large in players' minds.
The data help us to ask the right questions. Sometimes the reasons for a player's change in form are not so obvious. Consider Ederson's shotstopping of non-headers in open play since he came to the Premier League:
Something changed for Ederson at the dawn of the 2018-19 season. He had begun out as an average or slightly subpar shotstopper, but in 2018-19 his rating started rising and has stayed high ever since. Was it because Manchester City sold Joe Hart and Angus Gunn after handing Ederson a long-term contract? Or did he work on something in his technique over the summer? We're sure other GKs would like to know!
For now, we only publish our latest shooting and saving ratings, on a 0-99 scale for the player's position and your benchmark league. These ratings should be interpreted as a basis for expectations of players in their next matches – that is, how they're performing today, not how they performed two or three seasons ago. If you'd also like to see more of these time-series charts for other players, we can produce them on a bespoke basis. Just get in touch via the Contact link below. We look forward to hearing from you!
[Photo: Aleksandr Osipov]