Analytics Practice: Compare team performance fairly


German article here : Team Leistungen fair vergleichen


Data analytics is getting a bigger and bigger place in the sports world. Meanwhile, there is almost no club that does not want to build or expand its analytics skills. Besides clubs using data for scouting & match analysis, more and more sports formats are now integrating fixed analytics content/experts in their features. Whether in the press, TV formats or podcasts. Analytics is on fire. In Malaysia there is now even the first dedicated data analytics program for the local championship. In Switzerland, it won't be long before the first data analysis expert sits in a sports studio.


Let'start to get analytical


In one of our past articles, we discussed how to compare players' performances as fairly as possible. Because to assess quantity or volume based values (number of duels, dribblings, shots, passes etc.) per 90min is good but not always the best solution. A smarter adjustment per possession, ball touches, or passes makes the values and the comparison fairer and more meaningful.


Here again our blog about it: DataAnalytics practice: comparing players fairly


Today we do an example on team level and take you into the analysis process.


see the process - not the result

We want to understand how well teams are getting into the final third with their build-up play and how often they can touch the ball in the opposition penalty area.


There are only 6 Super League rounds played and therefore the values are not yet sufficiently reliable. Experience has shown that a value of 800 minutes or more is a sufficient quantity.

In our example, however, we are more concerned with the method than with the findings.


Therefore we use the metrics "Passes to final third p90" and "Touches in penalty box p90" .



Unadjusted values: Passes to final third & touches in penalty box


ability to reach final third and box (not adjusted)
ability to reach final third and box (not adjusted)

Maybe you take a moment to study the picture and the positioning of the teams. Does the picture match your perception?



St. Gallen, YB, Lugano and Basel often manage to get into the final third and touch the ball in the penalty area. Sion, Winterthur, Servette and Luzern show mediocre values in both dimensions. Zurich manages to get into the final third well, but then has trouble successfully penetrating the penalty area. The Grasshoppers don't seem to care about these statistics. They let the opponent have the ball due to a low possession rate and concentrate on the switching moments.


Of course, a large volume in these dimensions does not necessarily mean being successful. Of course, more touches in the penalty area increases the risk of scoring, but

a team can get into the box 30 times in a game without scoring a single goal.

The level of the values says so (unadjusted), much more about the playing philosophy of a team than about the quality.


In the picture above we have only compared the values or the volume of the different teams. We did not take into account how much time or how much possession the team had to execute its actions.


After six rounds of the current swiss Super League championship, Basel has the highest possession value with 57.9% and Winterthur has the lowest with 40%. If we extrapolate that to an effective net playing time of 60 minutes, Basel has 35 min the ball and Winterthur but only 25 minutes. So Basel has 10 min more or almost +30% more time to execute its actions. On the other hand, Winterthur had 30% more time to execute their defensive actions.


It would not be fair to compare the volume based performances of of both teams without an adjustment.

Thus, in a second step, we will now adjust the teams' values to the amount of possession they have. We adjust all values to 50% possession and arrive at the following result.





Possession adjusted values


ability to reach final third and box (adjusted)
ability to reach final third and box (adjusted)


The values of the teams are now adjusted in a fairer way and therefore more comparable and meaningful. We now measure things with equal cubits and are able to make a fairer assessment.

We now simulate what the values would be for teams with the same amount of possession, while maintaining the same quality. Thus, we are closer to the truth and the statement has changed from a pure question of style to a question of performance.



Without value adjustment, players from top teams and the top teams offensively are weighted better because they have more of the ball. The same is true for defense in reverse.


A (smart) adjustment often puts players from weaker teams in the deserved spotlight and can really make them shine.



Also in our chart there have been exciting shifts, but also confirmations, which will hopefully surprise you as well.



We are happy to leave the findings to your interpretation and leave them uncommented.









Hey, we're still here, as we've decided that the analysis isn't over yet.

We'd like to dig further to perhaps get even better and deeper insights.



The stronger the correlation, the stronger the insight.


We took team ball possession as the unit of adjustment, and that's fine. But is that really the best unit available? A team can have a lot of possession in its own half and the many passes in the build-up to the game have little connection or correlation to the touches of the ball in the penalty area. So I think about what unit has even better correlation. What could correlate even better.


I'll try this: for the "passes to final third", I'll stick with possession. Because you can say, the more time in the buildup of the game, the more opportunities I have to get into the opponent's third. A direct dependency is there.


With the "Touches in the box" I use the number of passes into the last third for the adjustment. This means that I have eliminated the passes of the buildup game in the first two zones from the context. The more passes into the last third, the more chances of touching the ball in the penalty area. This has a stronger correlation than pure possession. These two values have a stronger relation and dependence on each other.




And already that stands next diagram.


ability to reach final third and box (advanced adjusted)
ability to reach final third and box (advanced adjusted)


And again, the picture and the statement has changed somewhat.

How do you now interpret the values and the performance of the teams?

What are the insights or valuable new questions?


Now one could object: That for the Adjustment the runs with the ball, which also bring the ball into the last third, are missing. Or that the touches in the penalty area also come from standards and that this distorts the statement.


That is absolutely correct and would certainly change the picture again somewhat. But the runs into the final third and the touches in the penalty area from play only, are not included in our available dataset. We used the best available unit and compared all teams equally.


We used statistical data from Wyscout for our analysis. We would have much greater possibilities with the much more detailed event data. For a club, the purchase and analysis of event data (3'000 events per game, including start and end coordinates of each action) opens new and great potentials and possibilities for game analysis. Event data can be used to answer many questions, develop detailed analyses and validate hypotheses.


In data analysis there is no absolute truth.

We try to get closer to the truth piece by piece with smart questions and with the validation of hypotheses. We start with a question and try to gain information from the data and insights from the information. The insights in turn lead to valuable questions that we try to answer.


Football is our great passion and countless exciting questions and answers await us.



An expert manages to enter into a profound dialogue with the topic so that both sides always have something to say.


Data analytics is a bit like treasure hunting. Theory and knowledge paired with courage in action.

"It's not enough for an football club to just buy in data and analytics tools. That's what everyone does - it keeps you one in the crowd. Invest in data, software AND #DataAnalytics knowhow."



Did you like the quality of our post? Then reward us with your "credits" and like and share this post within your social network. Thank you.



footballytics – mit smart data die besseren Entscheidungen treffen

We combine the competencies of tactics, scouting and data analytics and advise and support clubs in interpreting and using data to make better decisions in scouting and match analysis.

 

www.footballytics.ch About Data Analytics in football - improve the game - change the ǝɯɐƃ

Share this Post

0 Kommentare

Aktuelle Beiträge

Alle ansehen