Visualizing a Team's Goal Chances in Soccer from Attacking Events: A Bayesian Inference Approach

Area of Science:

• Sports Analytics
• Statistical Modeling
• Soccer Performance Analysis

Background:

• Traditional soccer analysis often lacks detailed insights into chance creation and player roles.
• Understanding the spatial and player-specific dynamics of attacking plays is crucial for performance evaluation.

Purpose of the Study:

• To develop a method for inferring the number and nature of scoring chances in soccer using only attacking event data.
• To quantify team abilities in chance creation and identify key player contributions and their on-pitch locations.
• To visualize and compare player attacking styles based on assist and chance locations.

Main Methods:

• Utilized a Bayesian inference framework with a Poisson model to estimate chance occurrences and team abilities.
• Employed a Gaussian mixture model to analyze the spatial distribution of assists and chances.
• Applied the developed scheme to the 2016/2017 English Premier League dataset.

Main Results:

• Successfully inferred team abilities to create scoring chances from attacking events.
• Identified specific pitch areas where players most frequently make assists and take chances.
• Visualized player differences in attacking play, highlighting key impact zones.

Conclusions:

• The proposed method offers a novel, interpretable approach to soccer analytics using solely attacking event data.
• This technique provides deeper insights into team offensive capabilities and individual player impact on chance creation.
• The findings enable a more nuanced understanding of attacking strategies and player performance in professional soccer.