Recently, I received access to detailed data about the 2017 Brazilian soccer League (also known as Brasileirão). Therefore, the next few posts will most likely be more dedicated to the Brasileirão, it is important to say that I accept suggestions and ideas for future studies! Today I will talk a little about the tournament’s top scorers using some metrics and concepts present in previous texts.
Assumptions:
We assume that it is possible to measure and evaluate the differences between some of the Brasilerão’s top scorers with a metric called expected goals (abbreviated to xG). Expected goals is a metric that has gained a lot of popularity among different football analysts. Basically, this metric assigns a value between 0 and 1 for each shot per the probability of that attempt being scored. For example, a shot from outside the box with the weak foot (shot 1) is less likely to result in a goal compared to a shot inside the small box with an open net (shot 2), therefore shot 1 has a lower expected goal value than shot 2. With xG we can better understand the quality of the chances created and conceded by each team throughout the match.
xG is very interesting data because it provides extra information about the shots made by a player. The idea is that at the end of the match you can add the xG value of all the shots attempted by a player and say that “the player was expected to score Y goals.” In other words, the xG allows us to evaluate whether a player could take advantage of the opportunities he had. It is also possible to understand how unlikely some goals are.
Methodology:
I used the Stratabet data on the Brazilian League and created a “xG (expected goals) scorer ranking”. Basically, I added the xG values of each chance and shot attempted by each player to create this list. Then, I placed this ranking next to the “traditional” top scorer list (in which only the goals scored by each player are considered). After that, I calculated the difference between the expected goals and the goals scored by each player, that is, I made the following subtraction EXPECTED GOALS – GOALS SCORED. The idea is to be able to know how far from the expected is the performance of the Brasileirão’s shooters.
It is important to note that, when writing this article, the data had last been updated after the 14th week of the tournament, which means it does not include the goals and chances of last Wednesday’s and Thursday’s games (July 19 and 20). Once the data is updated I plan to post new versions of the xG ranking.
Results:
Here is the table that brings, side by side, the “usual” top scorer list and the xG rankings:
We can see that many names have close numbers in both lists. A good example is Henrique Dourado who has a xG of 8.79 and scored 9 goals, that is, the Fluminense striker was expected to score 8.79 goals and he scored 9. Another example is Lucca, the Ponte Preta attacker, who has a xG of 6.53 and 7 goals scored. The number of goals made by these players is in accordance with the quantity and quality of chances that have had so far in the Brazilian league, they did not waste very clear chances, but also did not score too improbable goals. In other words, players who have an xG close to the number of goals they have made are taking advantage of the opportunities throughout the game in the expected way.
However, some names have a big difference between their xG (expected goals) and the actual amount goals scored. Let’s first analyze those who have an xG greater than their number of goals:
The image above displays the players who have the biggest difference between the xG and the amount of goal, which means they are making fewer goals than expected. This difference between the xG and the goals scored can be understood in two different ways. The first is that these players are simply experiencing a period of “bad luck”. They shoot many times and create good opportunities, but sometimes the ball just doesn’t find the back of the net. The second possible prespective is that they have been inefficient. It can be understood that players like Pratto, Luan and Everaldo, even though appear among the league’s top scorers, are wasting too many chances and “lacking” in terms of goals.
Now let’s look at the other extreme, players with a number of goals significantly higher than their xG:
The players present in the chart above have a significantly higher number of goals than their xG, it can be said that they are scoring more goals than expected when taking into consideration the quantity and quality of their shots. Again, the difference between the number of goals and the xG can be understood in 2 ways. On the one hand, you can see these players as “lucky”, making improbable goals and participating in matches in which everything seems to work out right. On the other hand, it can be said that the numbers of these players are not related to luck, but to efficiency. Following this theory, strikers like Copete, Everton and Jô have been managed to be extremely accurate and convert many their shots into goals.
The debate between luck vs. efficiency in soccer is very interesting and subjective. It is not easy to understand exactly how competent or lucky a player is. Even after having very impressive numbers over the course of a year, it is hard to say for sure if the player really is more efficiente and competent than expected or if he just had a lucky season. I believe it is easier to understand luck and merit when looking at a more macro level and analyzing the team as a unit. I intend to do this in a future post.
Observations:
- When writing this article, the data had last been updated after the 14th week of the tournament
- In this text, I gave a focused on the shots, however a similar analysis can be done with the assists.
- Missed penalties can have a great weight in the xG data and are an extra ingredient in the discussion about luck vs. competence. The question always remains, was the penalty missed due to the penalty takers lack of skill or just statistically unfortunate bad luck?
This article was written with the aid of StrataData, which is property of Stratagem Technologies. StrataData powers the StrataBet Sports Trading Platform, in addition to StrataBet Premium Recommendations.
Algolritmo's founder. I have a bachelor's in Computer Science and a master's in Analytics. My goal is to bring a new perspective into Brazilian football. I'm particularly interested in communicating complex ideas through simple data visualizations.
I graduated in Computer Science and Business Administration at the University of Southern California and got a masters degree in Analytics at the same institution. I have worked as a Data Science intern at companies such as Facebook, Itaú and Looqbox.
Ola Rafael .
Muito bacana esse trabalho e com mercado muito grande a ser explorado !!
Att.
Muito obrigado Paulo! Espero poder ser um pioneiro nesse mercado!
Abraços