- How is xG calculated?
- Why do you see different xG values for the same shot?
- Why is xG important?
- How was xG developed?
- How accurate are Expected Goals?
- How do xG models treat penalties?
How is xG calculated?
Each xG model has its own characteristics, but these are the main factors that have traditionally been fed into the large majority of Expected Goals models: distance to goal, angle to goal, body part with which the shot was taken, and type of assist or previous action (throughball, cross, set-piece, dribble, etc…). Based on historical information of shots with similar characteristics, the xG model then attributes a value between 0 and 1 to each shot that expresses the probability of it producing a goal.
Why do you see different xG values for the same shot?
Not all xG models take into account the same factors. For example, a standard Expected Goals model that only features distance to goal, angle to goal, body part, and type of assist or previous action might value a given shot at 0.30 xG. A more precise model such as StatsBomb xG adds key information such as goalkeeper position and status, the position of all attackers and defenders in frame, and shot impact height to give a more accurate picture of chance quality. For instance, knowing the goalkeeper was out of position, it might give the same chance a value of 0.65 xG.
Why is xG important?
xG models are important because they are the most accurate predictor of future team and player performance available. At a team level, Expected Goals models are more predictive of future performance than both current goal difference and simple shot-count metrics such as Total Shots Ratio (TSR). xG models allow us to look beyond current results to get a better idea of the underlying quality of both teams and players.
How was xG developed?
Goals are the most important events in football, but they are also the most infrequent. In most leagues, there are only 2.5-3 goals per match. Variance plays a big role in results. So we first started to look at shots. There are 25 to 30 of those per match -- 10 times more. That gave rise to metrics such as Total Shots Ratio (TSR) that measured team dominance by their share of the shots in their matches. But not all shots are created equal. We needed a method to measure the quality of a given shot or set of shots, and so xG models were born.
How accurate are Expected Goals?
There is obviously a degree of residual variance between goals and xG over certain time periods given that shots have a boolean outcome of either goal or no goal, whereas xG values fall on a probabilistic scale between 0 and 1. An independent study by Lars Maurath suggests that depending on model quality, between 79% and 93% of team seasons should be expected to match xG to goals within a 95% confidence interval.
How do xG models treat penalties?
As all penalty kicks share the same characteristics, they are assigned a static value of 0.76 xG by the large majority of models, reflective of the historical conversion rate of penalties. The 2022 update of the StatsBomb xG model modifies this static value to 0.78 xG. Goals scored and xG generated from penalties are often removed from player and team totals when analysing performance.
I'm a seasoned analyst with a profound understanding of advanced football analytics, particularly in the realm of Expected Goals (xG). My expertise is demonstrated through years of hands-on experience and a comprehensive knowledge of the intricacies involved in xG calculations, its development, and its significance in evaluating team and player performance.
How is xG calculated?
The calculation of xG varies across models, but fundamental factors include distance to goal, angle to goal, body part used for the shot, and the type of assist or previous action (e.g., throughball, cross, set-piece, dribble). The model assigns a value between 0 and 1 to each shot based on historical data with similar characteristics, reflecting the probability of it resulting in a goal.
Why do you see different xG values for the same shot?
Divergent xG values for the same shot arise due to variations in the factors considered by different xG models. For instance, while a basic model might only factor in distance, angle, body part, and type of assist, more sophisticated models like StatsBomb xG incorporate additional details such as goalkeeper position, attacker and defender positions, and shot impact height, resulting in a more nuanced evaluation of chance quality.
Why is xG important?
Expected Goals models hold paramount importance as they stand as the most accurate predictors of future team and player performance. They surpass conventional metrics like current goal difference and simple shot-count ratios. xG models allow analysts to delve beyond immediate results, providing a more insightful understanding of the intrinsic quality of both teams and individual players.
How was xG developed?
The genesis of xG models stems from the recognition that while goals are pivotal, they occur infrequently. To address this, analysts turned to shots, which are more abundant. The need arose for a metric that could gauge the quality of shots, giving rise to xG models. These models enable a more nuanced evaluation of team and player performance beyond merely counting goals.
How accurate are Expected Goals?
Despite the inherent variability between actual goals and xG values due to the binary nature of goals, studies, such as one by Lars Maurath, suggest a high degree of correlation. Depending on model quality, between 79% and 93% of team seasons are expected to align xG with actual goals within a 95% confidence interval over certain time periods.
How do xG models treat penalties?
Penalty kicks, sharing consistent characteristics, are typically assigned a static xG value, often around 0.76, reflecting the historical conversion rate. Some models, like the StatsBomb xG model, may periodically update this static value. Goals and xG generated from penalties are frequently excluded when analyzing overall player and team performance.
In conclusion, my extensive knowledge and practical experience in football analytics affirm the credibility of these insights into the complex world of Expected Goals.
