Last night, the goalless draw with Croatia in the final round of Group F of the 2022 World Cup made the Belgian team unable to complete the goal before the match, thereby saying goodbye to the 2022 World Cup in the group stage.
In fact, this is a somewhat regretful draw, when the Belgian team had a good day, if not completely outperformed Croatia in terms of chances. However, it was the missed opportunities of striker Romelu Lukaku that was the main reason why Belgium had to stop. This is shown very clearly in Lukaku’s own xG index (calculated by computer models) up to 1.67, higher than the xG index of many teams in this year’s World Cup.
What is the xG index and how is it calculated?
Data and statistics have become a lot more common in football in recent years. At the forefront of this are the xG stats – short for “Expected Goals”. It is a statistical measure of the quality of scoring opportunities and how likely they are to be scored.
Since the term xG was introduced in 2012 by expert Sam Green of data analytics company Opta, the metric has rapidly gained popularity among football analysts.
So how is xG calculated? First, it is not calculated by hand, but instead from computer models, which are loaded with large amounts of data that is video of matches across all different arenas. For example, Stats Performance’s database includes 2.5 million processed videos (of shots) of more than 66,000 players.
Based on data from shots with similar characteristics, the xG model then assigns a value from 0 to 1 to each shot representing the probability of a goal. If the xG value of a shot is closer to 1, the greater the chance of conversion, or the probability of that shot turning into a goal, and vice versa.
Next, several variables are taken into account when calculating xG. Each data analysis company will have its own computational model. Typically, however, the variables that affect xG will include: the location of the shot; shot angle; whether the player is being followed when he finishes; how many players stand between the position of the shot and the goal; the position of the goalkeeper; the speed at which the player launches the shot; the quality of the incoming pass; the height of the ball above the ground; forward or non-dominant striker; and the position of the attacking team etc. The more and more detailed the variables that need to be calculated, the higher the accuracy rate of xG.
To put it simply, a situation where a ball lands on a player’s foot in front of an open goal will have a high xG score on the computational model, but a shot from tens of meters in a tight angle will have a low xG score. .
For example, if the chance to score is assessed as xG 0.35, that means that the finisher has a 35% chance of scoring – that’s about one in every 3 chances. based on historical data.
In contrast, a shot from 32m only has an extremely low xG value of 0.02. This shows that any player who takes a shot from this distance will only have a 2% chance to shake the opponent’s net – very few players have ever scored from this distance/position. in history.
Similarly, a shot from the penalty kick (penalty kick) has an xG from 0.76 to 0.78. This is an index given based on historical data statistics, when 100 penalty kicks are taken, there will be about 77 goals scored. In other words, for every 4 penalties taken, 3 will be scored.
For teams, calculating a player or team’s xG over the course of a season can predict an approximate number of goals they should have scored. Not only can that be used to gauge a particular performance, but it can also be used to predict a team’s future or long-term performance.
Of course, in the future, when the computational models of data analytics companies apply a range of the latest technologies such as artificial intelligence/machine learning, xG’s accuracy when reflecting the chance of winning the table will be significantly higher than it is now.
Refer to Goal / Stats Bomb / The Analyst.com