XG

XG, short for expected goals, is a statistical metric used in football analytics to quantify the quality of goal-scoring chances and estimate how many goals a team or player would be expected to score based on shot opportunities.

Calculation and inputs: For each shot, a probability of scoring is assigned according to a model trained on historical data. The shot’s xG equals this probability, and a team’s or match’s xG is the sum of the xG values for all shots. Models use various features, including shot location (distance and angle), shot type (foot, head, volley), whether it occurred in open play or from a set piece, build-up context (assists and previous passes), and sometimes time or defensive pressure. Different providers calibrate and weigh features differently.

Interpretation and uses: xG measures shot quality rather than finishing skill. It is commonly compared with actual goals to assess finishing efficiency and to judge how well a team or player creates high‑quality chances. Applications include evaluating team performance and tactics, scouting and recruitment, player development, and forecasting future results or informing fantasy sports decisions.

Variants and availability: Multiple models exist, with providers such as Opta, StatsBomb, and Understat offering xG data at team and player levels. Differences among models arise from input features, data timing, and calibration, which can affect absolute values and year-to-year comparisons.

Limitations: xG depends on data quality and model design and may not capture goalkeeper actions, shot context after release, or dynamic pressure and rebounds precisely. It should be used alongside other metrics for a complete assessment and interpreted within the context of the league and dataset.