Identifying Attacking Types with Data: The Key to Optimize Offenses & Defenses
Any type of analysis consists in first identifying general patterns that lead us to the more important aspects, and then take a deeper dive and find an improvement opportunity.
For example, to analyze a team’s strengths and weaknesses, the first thing we do is see how well the team is attacking and defending in different situations: set pieces, counter attacks, possession play, etc.
For a video analyst it’s relatively easy to identify game situations. Even if there are no manual definitions, we can recognize a possession pattern just by looking at it.
This looks really simple. But the problem is when this manual task is used to analyze many teams’ full seasons or watch all games from players to be scouted.
This is when identifying game situation with data becomes really helpful. By doing this, we are able to reduce a lot coaches’ time by finding general patterns and filtering situations for them to get the most important conclusions with video tools.
Let’s first look at the different types of attack and then see some examples on how we could use it to make teams’ diagnostics.
Let’s explain briefly each game phase taken into consideration.
1. Set Pieces
This is a possession that comes from a free kick or corner kick.
2. Counter Attack
A counter attack possession is a fast one in which the attacking team has recovered the ball before 60% of the field (being 0% their own goal).
3. Fast attack
This kind of possession is very similar to a counter attack but with the condition that the ball was recovered after 60% of the field.
4. Direct Attack
This is a possession that starts in the team’s own half and is not as fast to be a counterattack or starts from a dead ball (throw-in, free kick, goal kick). The team must also make at most 1 touch between half and 67% of the field and at most 3 touches in the 3rd third.
5. Possession attack
By definition, it is an attack that does not fit into the other phases. It means that the situation is built on possession.
From data to reality
Now that we’ve identified attack types, we are able to analyze teams’ characteristics to get some insights.
In these diagnostics we will use the following metrics:
- Expected Goals (xG) percentage from each type of attack to analyze teams’ offensive and defensive trends.
- Expected Goals (xG) for every 100 possessions that end up in a shot to measure the danger created from each type of attack.
The following analysis correspond to the first 8 weeks from Argentinean Superliga 2019–20.
1. San Lorenzo’s Possession Game
San Lorenzo chooses to attack mostly by building possession plays. The dependance on this kind of situations can be a problem in case the team needs a plan B.
2. Fast Attack: Velez’s Lethal Weapon
Velez is really good at generating fast velocity attacks. They are the ones that get the most value from this kind of situations. To optimize this kind of attacks, the next step could be analyzing pressure and recovery zones.
3. Racing’s Dependance on Set Pieces
The majority of Racing’s attacks come from set pieces actions. Even though this kind of situation is important, it could be a good idea to investigate why they are unable to generate more chances by other means.
4. Estudiantes’ Counter Attack Problem
Estudiantes suffers a lot from counter attacks. The combination of getting the majority of they xG conceded in this way and it being the most xG/100 pos phase makes it really dangerous for them. One next analysis could include losses zones and patterns.
After this type of diagnostics, we can guide our analysis towards something deeper, depending of what we are trying to solve or optimize. This following analysis could include:
- Starting possession zones.
- Shots & assists zones.
- Players involved in each kind of actions.
Once we are done with this exhaustive data analysis, now the video analyst can focus only in the improvement opportunity without watching and analyzing manually all chances.
On the other hand, this kind of information can be really helpful in scouting. For example, when we want to look for an offensive player, we can filter not only by their xG, but now we can search by their values in playing styles that are similars what we want in our team.