Eduin Boater Latimer analyses how finishers fare season-on-season and how much moving teams can affect the ability to out-perform expected goals
Mateo Retegui is having a huge year. The Argentine striker has led Atalanta’s Serie A title charge this year, scoring 18 goals (excluding penalties) in only 22 matches. His impressive performances have put Atalanta in the title race and are rumoured to attract attention from potential buyers, such as Arsenal, PSG, and Manchester United.
There are two reasons he’s scored so many goals this season. First, he’s taken lots of shots from good locations. Expected goals (xG) are a measure of how likely a shot is to score based on its location and other characteristics. Players with high expected goals take a lot of shots in good locations. Retegui has generated 9.4 non-penalty expected goals (0.64 per 90 minutes played), the second most of any player in Serie A this year. Second, he has finished well. He has scored 18 goals from just 9.4 expected goals.
His recent performances raise an important question for both Atalanta fans and interested potential buyers. Can he sustain these performances. Will he be able to continue to put up high expected goals and continue to overperform those expected goals?
Mateo Retegui’s Statistics since 2019-20
| Season | Age | Team | Minutes | Matches | Shots | Expected goals (xg) | xg per 90 | Goals | Goals/ xg |
| 2019-2020 | 20 | Estudiantes | 991 | 18 | 26 | 3.3 | 0.30 | 3 | 0.91 |
| 2021 | 21 | Talleres | 1,038 | 24 | 35 | 4.9 | 0.42 | 4 | 0.82 |
| 2022 | 22 | Tigre | 2,231 | 27 | 91 | 9.8 | 0.40 | 13 | 1.33 |
| 2023 | 23 | Tigre | 1,790 | 21 | 79 | 7.6 | 0.38 | 9 | 1.18 |
| 2023-2024 | 24 | Genoa | 2,216 | 29 | 62 | 5 | 0.20 | 6 | 1.20 |
| 2024-2025 | 25 | Atalanta | 1,316 | 22 | 62 | 9.4 | 0.64 | 18 | 1.91 |
| Total since 2019-20 | 9,582 | 141 | 355 | 40 | 0.38 | 53 | 1.33 |
Source: Opta via FBRef. Note: Both goals and expected goals exclude penalties.
The persistence of expected goals
First, let’s look at the persistence of expected goals. We can do this by testing if forwards with high expected goals in one period are more likely to have high expected goals in the next period. Initially, when looking at consecutive periods of 5 or 10 matches, there is a relationship, but it is weak. For longer periods the relationship is stronger. When looking at consecutive periods of 30 matches or longer, the slope of the curve is around 0.6. We’ll call this the persistence factor from now on. This means that if a player’s xG per 90 is 0.1 per 90 above average (among forwards in the big six leagues) in the first period, we would expect it to be 0.06 above average in the second period. So, xG per 90 is strongly persistent: if a player has high xG in one season, they tend to have high xG in the next season.
Expected goals measures a player’s ability to get into good position to take shots, but this clearly depends on other players on their team. If a forward plays with creative players who are good at finding them in good positions, they will be more likely to get high expected goals. So, we can test how important a players’ teammates are by checking if xG is persistent for players who switch teams as well as players who stay on the same team. The chart below compares all forwards (on the left) with forwards who play for different teams in the first and second period (on the right). It shows that xG in the first period strongly predicts xG in the second period, even for players who switch clubs. The relationship is a bit weaker for players who move – a persistence factor of around 0.4 – but still strong. This suggests that a key part of a forward’s xG is due to their own ability rather than the ability of their teammates.
So, we’ve found the ability to get into good positions is strongly persistent and mostly reflects the ability of an individual forward. This bodes well for Retegui, he has registered 0.38 expected goals per 90 over his career and 0.64 per 90 in the last season. Both figures are well above the average for forwards of 0.19 per 90, so he is likely to continue getting good goalscoring opportunities. But what about his clinical finishing, how confident can we be that it will persist?
The persistence of finishing: goals relative to expected goals
We can carry out a similar test for persistence of finishing. We can measure finishing using goals divided by xG (goals/xG): how many goals a player scores for a given quality of opportunities. The chart below shows the persistence of goals divided by xG (goals/xG) for players across different numbers of shots. We can see that across a small number of shots, there is almost no relationship. Goals/xG is a noisy measure of finishing ability and in small shot samples it is essentially random. When looking at larger samples of 60 shots or more, there is a clear, if weak, relationship. The persistence factor is around 0.1. The average goals/xG ratio is slightly below 1. So, if a player strongly over-performs their xG in one period with a goals/xG ratio of 2, we would expect them to over-perform their xG in the following period but typically by a much smaller amount with a goals/xG ratio of around 1.1.
Analysts tend to talk about players over-performing their expected goals in two extreme ways. One group tend to talk about any over-performance of expected goals as just random noise, not at all related to a player’s finishing ability. Another group talk about overperformance of expected goals as if it is a strong signal of a player being a clinical finisher. We can see from the data that the truth lies between these two viewpoints. Some players are better finishers than others but there is also a lot of random variation in goals relative to expected goals. In small number of shots, the random variation dominates and goals/xG is basically random. Over large enough samples, a player’s goals/xG ratio does provide a weak signal of how good a finisher they are.
What does this mean for Retegui? This season he has a goals/xG ratio of 1.9 off 62 shots. Looking at this alone, we might expect his goals/xG ratio in future to regress back towards the mean and be closer to 1.09 in future. If we look over his whole career, he has a goals/xG ratio of 1.2. This is another signal that he is a good finisher and will likely continue to over-perform his xG, but that he is unlikely to continue this season’s excellent finishing. Any scouts evaluating Retegui should see his finishing as a bonus. His strongest selling point is his ability to get shots off in good positions, which I have shown is strongly persistent across seasons.
Key takeaways for scouts more broadly:
- Expected goals is strongly persistent amongst forwards. If a forward has an expected goals per 90 0.1 above the average for forwards in the big 5 leagues in one full season, we would expect them to keep 60% of that performance and be 0.06 above average in the next season.
- Expected goals is even persistent amongst forwards who move teams. This suggests that an important component of a player’s expected goals is their own ability separate of the ability of their teammates. If a forward has an expected goals per 90 0.1 above average in one full season, we would expect them to keep 40% of that performance and be 0.04 above average in the next season if they switch teams.
- Finishing (measured by goals divided by expected goals) is weakly persistent over long periods. Some players are consistently better finishers, but even over periods of 60 or more shots, most of a player’s measured finishing performance is unlikely to persist. If a player has a goals/xG ratio 1 unit above the average for players in the big 6 leagues over 60 shots, we would expect them to keep 10% of that over-performance and perform 0.1 over average in the next 60 shots.
Header image copyright IMAGO / IPA Sport / Davide Casentini
