AstaFut is a hobby Data Science group with a few colleagues, which uses Big Data Analytics to optimize soccer teams' performance. We believe that applying analytics to large amounts of unexplored data produced by teams can yield powerful results to players' and coaches' decision-making process.

We developed tools - particularly computational models and machine learning models - to analyze this data and highlight areas of improvement for players' on-pitch actions, and for teams' off-pitch planning. We also focus on data-visualization frameworks that aid in conveying our findings in a simple and intuitive manner, such that it is clear to anyone, from soccer players to data scientists.

An example model we developed is our 'expected goals' model, commonly referred to as 'xG'. This custom-made machine learning model allows us to predict the percent change of a shot resulting in a goal given a series of features: distance from goal, angle from goal, preferred foot, header, among others. From this model we are also able to infer how many goals a player is expected to score in a season given their current shooting strategy and provided opportunities. 

Our work is currently exclusively focused on the Brazilian soccer league, and therefore most of our work is developed in Portuguese

See: and

The xG visualization tolls above allow us to intuitively understand the shooting patterns of a given player and its effectiveness.

The size of the circle represents the shot's xG, i.e. the bigger the circle the higher the probability of scoring from that shot. The color green represents a goal, yellow is a shot on the post, blue is a shot saved by the opposing keeper, and and red is an off-target shot.

This knowledge can be used by his team to improve his offensive decision-making, but also defensively by opposing teams to neutralize this player

Analysis on the impacts of two given features (shot's angle and distance from goal) on the xG output

Our radar plot framework allows for a quick comparison of players' performance and play-style

Analysis of the five players with the biggest under-performance in the Brazilian league based on how many goals they should have scored from the shooting opportunities they received