How AI detects and measures pressing
Pressing is the pressure a team puts on the opponent who has the ball. By eye it is hard to quantify, but with tracking data (the position of every player many times per second) an AI model can measure it.
A well-known metric is PPDA, which counts how many passes a team allows the opponent before attempting a defensive action. The lower it is, the more intense the pressing. With more detailed data, models can detect where the pressing starts, how many players take part and how long they take to close down the opponent.
They can also identify “triggers”: specific situations that make the team press, such as a backward pass or a poor first touch.
This is useful to analyse the opponent and assess your own team. But a number does not explain the why: low pressing may be a tactical decision, not a lack of intensity. That is why coaches combine these metrics with video.