In the first of a series of semi-regular articles on this site, we will use data analysis techniques to find undervalued but successful young players in their respective leagues. There are many leagues across Europe in which the title race will tend to be dominated by a small group of teams.
Unless you’re a fan of that league or have an unhealthy obsession with the game of Football Manager, you’re unlikely to know all the interesting young players who are in those leagues but don’t play for a team that will traditionally challenge for the title. There are, of course, exceptions to this rule and in the German Bundesliga, where Bayern Munich are in a league of their own, there is more mainstream media exposure and as such wider awareness of players and team of this league.
When working in football recruitment, we want to have a broad knowledge base when it comes to identifying and recruiting talent, but we are limited by time and resources. So how can we build a reliable network or information that we can seek to access?
The answer is relatively simple. We can rely on the data as a tool to allow us to identify players who may match our profiles in leagues around the world. The problem is, however, that the player data for these leagues will always tend to be skewed in favor of the teams that sit at the top of the table. If we look at data from the Bundesliga, for example, we can expect to see Bayern Munich players playing a significant role. If we look at the SPFL players like Rangers and Celtic will feature prominently in our data set and if we focus on the Austrian Bundesliga we will see a low number of Red Bull Salzburg players. While this is not necessarily a problem and there are workarounds such as manually excluding players from these clubs from our dataset or simply ignoring them when they appear, there are other techniques that we can use to identify top performers at non-difficult clubs. up the table.
However, this may mean relying on readily available datasets. In this article, we will use wyscout data with python for coding and tableau for some visualization work. In python and tableau we have options to add additional calculations to our initial dataset to give the data a bit more life, but we’ll get to that before.
In this article we will start by looking at Serbia’s top flight and we will show the use of data analysis methods to identify four young players, we focus on players aged 23 or younger, who are performing well until now this season.
While we’re using data, we’ll focus for the purposes of this article on some key metrics. We are interested in goals and expected goals, assists and expected assists, progressive runs and progressive passes. Instead of just using the raw numbers for these metrics (which will give us a list of players who are under contract with the strongest teams in the league), we’ll use calculated fields in Tableau to apply a formula for give us the % a player is involved in each metric for his team. So if we have a striker who scored 10 goals but his team only scored 20 goals, he will have been responsible for 50% of the team’s total goals. This is compared to a player who has 10 goals but whose team has 50 goals as he is responsible for 20% of the team’s total goals. Doing this with all the metrics we are interested in will begin to show us which players are playing for smaller teams but are extremely effective for those teams.
First, we’ll be looking at goal contributions (goals per 90 + assists per 90) and the first thing to note is that we have to go down to 17th on our list before we find a player at Partizan or Crvena Zvezda. This is partly a result of the dataset I’m using as we’re only looking at players 23 or younger and despite their reputation for youth development, they have older squads this season. But it also begins to show us that our method works.
The top players in goal contributions both stand out, with Milos Pantovic (19, Vozdovac) and Uros Milanvanovic (21, Radnik) responsible for 14.63% and 13% of their team’s goal. contributions respectively.
We have to add at this point that this is only part of the process of identifying and recruiting players and then you need to take the players from the leaderboard that you are interested in and add them to a workflow to get started watching videos of the players in action.
Next, we’ll look at contributions to expected goals (expected goals per 90 + expected assists per 90)
Interestingly, Uros Milovanovic still tops this ranking, but this time the top spot is taken by 19-year-old forward Stefan Mitrovic of Radnicki Nis with a 13.76% contribution to his team’s expected goal contributions . This time we have to fight our way up to 13th place to find a player for one of the top two teams in the country.
Finally, for this section, let’s look at each player’s % contribution to their team’s progressive actions (progressive passes + progressive runs) because you can see this time the image changes position. The first two rankings we shared were usually populated with attacking players while this time we get defenders, full-backs and central midfielders. This is usually because players in these positions generate higher results for the metrics we are looking at. Forwards are more likely to score and centre-backs or full-backs are more likely to advance the ball. When I did this for a club, I always added position filters so you could easily identify outliers in positions you might not expect.
The likes of Filip Backulja (19 and on loan to Metalac from Juniors of Austria) who has 13.52% contribution to his team’s progressive stocks and Aleksa Damjanac (23 at Radnicki 1923) who has 12.57% contribution looks interesting but my pick is actually a midfielder in Stefan Purtic who is a 23-year-old central midfielder at Vozvodac.
#1 Milos Pantovic, 19, Winger, Vozvodac and Serbia
For these pizza charts, we take data from Wyscout and run it through python code to convert it to percentile data. This ranks each player in each metric against other players in the same position. in the same league from 1 to 100. The higher the value, the better the player is in this area compared to his peers. This percentile data is then taken and run through other code to produce the pizza chart above. If you want to make your own, the credit for the code is at the bottom of the table, but feel free to contact us for more information.
Pantovic is an attacking wide player who performs well in all possession metrics, but especially carrying plays and progressive plays. He also performs well in terms of dangerous passes and goal contributions.
#2 Stefan Mitrovic, 19, Winger, Radnicki Nis and Serbia
Stefan Mitrovic is a slightly different case in that he is a more traditional wide forward who thrives when carrying the ball but also provides above average outings (average is 50%) for goal contributions , shots per 90, shots on target percentage and hits. The area. He is a more effective player in the final third than to advance the ball. Interestingly, his positioning production is also very high in the defensive section.
#3 Uros Milovanovic, 23, Forward, Radnik and Serbia
As you would expect from a striker, the pizza board takes on a different shape compared to the two wingers above. Milovanovic is superb when it comes to goal contributions, expected goal contributions, shots per 90, shots on target percentage and touches in the box. He presents himself as a striker who still has development to come and with four caps for Serbia at U21 level, he has clear potential.
#4 Stefan Purtic, 23, central midfielder, Vozdovac and Serbia
Finally, let’s take a look at the only central midfielder in the group of players we’re interested in. Once again the shape of the pizza graph is starting to change with much more emphasis on the possession side of the game. His performance for dangerous passes and progressive actions is extremely interesting and as you can see from of his high performance for the passes received, he is a player who plays a lot with the ball for his team. His defensive performance for positioning also shows us that he can work against the ball in the defensive phase of the game.
So there we have it. Four Serbian players aged 23 or under and playing for teams that are generally not well known or well scouted. By using data and applying various analytical techniques, we can start to really dig deeper into the level of talent in various countries.
There is so much depth of talent there. The trick is to find players who match what you are looking for in terms of player profile and playstyle.