Valuing Passes Beyond Completion: A Look at Cross Field Passes
Which components of a pass gives teams the best chance of keeping possession
Most passing metrics in football data requires a pass to be completed in order for the metric to count or have value. However, football is very fluid and passes have more significance beyond whether or not they are completed.
This post looks at a very specific type of pass, a cross field pass, in order to try and evaluate what components of the pass itself and the receiver of the pass give a team a higher chance of keeping possession in their attacking half, regardless of if the pass was completed or not.
Defining a Cross Field Pass
For this purpose, a cross field pass was defined as pass that originated from a wide area up to the beginning of the wide half space area and ended in the opposite side of the field. The pass angle of the passes was equal to or less than 80 degrees.
A couple examples of what these passes look like can be seen in the image below.
Looking Beyond Pass Completion
As mentioned previously, passes are usually only valued when they are completed. However, this takes away from the idea behind a pass and certain functions that passes can serve.
Let’s take a look at an example from a Barcelona game against Athletic Bilbao in the 2020/21 season.
In the image below, the Athletic left back attempted a cross field pass to the opposite sided forward. The ball was overhit, and it went out of play.
Athletic set up well to defend Barcelona’s throw in and regained possession in their attacking half.
Thirty-five seconds after the pass went out of play, Athletic had regained possession and won a foul in their attacking third.
Even though the cross field pass was not completed, it allowed Athletic to move higher up the field and regain possession following the throw in.
Predicting Possession in the Attacking Half
Using a simple logistic regression model, I wanted to see which features best predicted possession in a team’s attacking half after an attempted cross field pass.
The data used for this model was the Statsbomb data from Barcelona Men’s games during the 2019/2020 & 2020/2021 seasons.
The biggest predictor of keeping possession in the attacking half after a cross field pass was the distance of the nearest opponent to the player receiving the pass.
The biggest predictors of not keeping possession in the attacking half after a cross field pass had to do with pass incompletion. Incomplete passes and passes that went out of play were the most significant features.
Conclusion
The model had a small effect size which means that the model has limited practical applications. This makes sense for various reasons.
Firstly, cross field passes are a specific kind of pass which narrows the scope.
Secondly, the video available for Barcelona games preceding the 2019-2020 season is scant which caused the data set to be on the smaller size. It would be interesting to see the results for a bigger sample size.
The work to collect the nearest opponents and nearest teammate distances was tedious, and resources such as the Statsbomb 360 Data would be a helpful tool in this area.
The purpose of this work was to initiate the idea that passes have a more holistic value on the team level.
This line of thought can open up other areas of interest. Teams could look at features of a pass that give a higher chance of possession in the attacking third or higher chances of creating shot opportunities.
Photo: Goal