Data Analysis with Game State
PlaymakerAI has introduced new features to simplify advanced football analysis. Starting this week, users can analyze football statistics with game state filtering. This powerful tool is easy to use and provides deep insights into how players or teams perform under different game scenarios, such as when leading or trailing in a match.
What is game state in football?
Game state is the balance in the game when it comes to goals in the game. If there is an equal score, both teams have 0 in game state. If team A is leading by 1-0, A will have game state 1 and B will have game state -1. Team A scores again and gets game state 2, while as team B gets game state -2.
Expected Goals per game in Allsvenskan when leading. Game State +
Expected Goals per game in Allsvenskan when being behind. Game State -
What can game state analysis reveal? What insights can it provide? Many have argued that a player’s mentality or ability to handle pressure during a game cannot be measured. PlaymakerAI challenges this notion, offering tools to analyze these aspects. How do a player’s statistics change when trailing in a game? How do they perform when leading, compared to when the score is level?
An analysis of Anton Salétros in Allsvenskan 2024 highlights his performance across different game states. The data reveals that Salétros is the most creative passer when AIK is behind. However, he wins the most balls in the offensive half when his team is leading. Overall, Anton achieves his highest ratings when AIK is losing, showcasing his ability to deliver his best performances when his team needs him the most.
Laurs Skjellerup's xG (expected goals) is shown in black, analyzed for situations when IFK Göteborg has an equal score and when they are leading. The yellow bars represent the total game units in these two different states. In Allsvenskan, an average game unit corresponds to 97 minutes.
Another approach to game state analysis is examining how teams are influenced by different game states. How do tactics and principles of play change, and what can this reveal about a team? Game state analysis offers a straightforward way to analyze opponents on a deeper level before a match. It also sheds light on how different tactical adjustments impact individual players and how they perform under these circumstances.
The spider chart above illustrates Djurgården's performance across three different game states during the Allsvenskan 2024 season. The team's performance varies significantly depending on the game state. For instance, Djurgården plays more directly when trailing compared to other situations. Notably, their xG (expected goals) is highest when leading, which may be influenced by the quality of their opponents in those situations or by the effectiveness of their attacking play when controlling a lead and playing without pressure.
How to do this in PlaymakerAI’s platform?
When generating your dataset we have added a a new bar that controls the game state.
PlaymakerAI offers 5 different stages of game state
-2, behind 2 goals or more in the game
-1, behind with one goal
0, equal score in the game
1, leading with one goal
2, leading with 2 goals or more.
It is possible to set the max min of the bar to the same value or you can set it at a range from -2 to 0.
Example screen shot of when creating a dataset. The game state settings can be done with both team and player aggregated dataset.
Are you working in the football or sports content business and interested in trying out PlaymakerAI’s platform nothing.
Book a demo and trial at info@playmaker.ai
Jesper Haglöf,
Data Scientist at PlaymakerAI