Ever since expected goals became the most mainstream advanced metric in soccer, much has been made of the difference between expected goals and actual goals. At first, it was believed that players who outperformed their expectation were “lucky” and would revert to the mean in future seasons. However, as we started to see more seasons of expected goals data, it was very clear that certain players were consistently outperforming their expected goals tallies.
From this, it became clear that certain players were better at finishing chances than others (a feature that is not currently incorporated in most expected goals models), and a lot of articles have come out examining the difference between XG and Goals. However, the metric that is often used to look at this difference is just a raw “Goals minus XG.” This is a good first start when trying to evaluate players finishing skills, but it lends itself to a bias of players who generate more chances, and have more opportunities to widen the gap between goals and expected goals.
As a result, I decided to try to quantify a different method to determine finishing skill. To do this, I wanted to calculate the probability that a player scored as many (or as few) goals as he did, given the opportunities he had. Therefore, a player who had 10 xG but in reality scored 15 goals is likely a better finisher than a player with 50 XG, and scored 56.
To go about solving this issue, I decided to take all players in the Understat data (up to November 24th of this season) who had more than 100 attempts on goal in the Big 6 European Leagues (England, Spain, Germany, Italy, France, Russia) since the 2014/15 season. This tallied 651 players. Then, I simulated all of their shot attempts 25,000 times based on their xG value, and calculated which players had the lowest probabilities of scoring as many goals as they did given the quality of their chances.
After running the simulations, we found that the following 30 players had the lowest probability of scoring as many (non-penalty) goals as they did.
This list sees many world class players such as such as Lionel Messi, Harry Kane, Antoine Griezmann, Eden Hazard, Philippe Coutinho and Mohamed Salah. However, there are some very interesting and unexpected players that appear on this top 30 list. The first name that sticks out is Serge Gnabry, who has struggled to get into the first team at Arsenal and Bayern Munich, but had very successful loan spells at Werder Bremen and Hoffenheim, where he scored 21 goals in 49 appearances.
A pair of Southampton strikers (Manolo Gabbiadini and Sofiane Boufal) who have signed at Saints since the rise of expected goals also appear on this list, which does make me a bit curious if this kind of analysis has featured in the recruitment policy on the South Coast.
It is also important to note the pair of Russian league strikers (Roman Eremenko and Fedor Smolov) in our top four. Their appearances towards the top of the list imply some level of top finishing ability, but could be due to other factors. Based on my research published last week using the Understat dataset, the Russian league has thrown up some weird results. This might due be due to faulty data coming from Opta, as hypothesized by this Russian journal article (after using Google Translate to translate the Russian). These Russian league strikers appearing at the top (as well as the appearance of other players on this list) could also be in part to poor goalkeeping as opposed to good finishing. Nonetheless, it is interesting to see these players at the top of our list alongside players such as Messi.
Next, let’s take a look at some of the worst finishers in Europe:
It is interesting to note that the p values for bad finishers increase much more rapidly than it does for the top 30 strikers. We find that Jesús Navas tops our list, who hilariously has already featured in a “worst finisher in the Premier League” article. Saido Berahino is an interesting player to appear on this list. Berahino was a surprise package for West Bromwich Albion in the first half of 2014/15, receiving an England call up and then an infamous failed transfer to Tottenham Hotspur. Since then, he has only scored 4 goals in 63 Premier League appearances, and failed to score in 28 PL appearances after transferring to Stoke City in January 2017.
Tottenham Hotspur supporters will be disappointed to see newly reborn midfielder Mousa Sissoko on this list, but perhaps if he had performed in par with his xG in his first two seasons in North London then this recent uptick in form wouldn’t appear so stark. Interesting to note, however, that he has yet to score in over 1000 minutes in any competition this season. Edin Džeko is also a surprising feature here, given that he has scored lots of goals in a Roma side that is often not quite good enough to overtake Juventus in Serie A. Overall, this list is much duller than the top finishers list.
I hope you enjoyed reading this, and I wish everyone a happy festive period and thank you for taking the time from your families and loved ones to read this. Later in the week, I’ll take a deeper dive into looking at which players are the best at finishing easy chances, as well as which players are the best at finishing harder chances.
If you have any questions for Andrew, please feel free to reach out to him by email at email@example.com or on Twitter @andrew_puopolo.