the xG or how sports betting can come close to science

by Mircea Panait

Even though some consider that sports betting also needs luck, as science advances, more and more mathematic models end up being adapted and applied. Alongside match analyses, statistical predictions, automation programs, the xG starts to become more and more attractive.

Still in an age of late-pioneering and new models of compiling primary data, the xG is an indicator  which associates to each occasion of scoring a success percentage based on data gathered and systematized in previous seasons.

xG comes from Expected Goals is defined as the answer to the question if a player should have scored in a clear position. The concept leaves Premier League, the most important championship in the world, and is already applied in the top English clubs.

Keep in mind that xG does not offer predictions about the future, bu analyzes the matches played thoroughly, and the obtained results are applied at the level of the championship. Even though no prognosis is offered, the data obtained through processing is extremely valuable and not easily accessible.

Regarding expected goals, in Premier League, 1 goal out of 9 kicks aimed at the goalpost. xG, expected goals, represents an indicator that takes into account each goal chance, trying to answer the questions should a player score on a certain goal occasion.

The maximum value for xG is 1, and the closer the number is to one, the higher the opportunity of scoring is. Therefore, if the chance is 0.5xG, then goals should have been registered in 50% of cases.

The image above is significant and has been used by to explicitly illustrate. Lamine Kone can be observed trying to score within the square at 6 m in the Sunderland – Everton match. In that case, the xG was 0.91, considered to be a huge opportunity. If he had shot 100 times from that position, he would have had to score 91 times.

As to better understand how the method works, those from Opta (the most important football statistical data provider) have analyzed over 300,000 kicks to the goalpost as to calculate the probability of scoring from a certain position and a certain game stage. A chart came up illustrating the following:

The factors considered by the analysts as to calculated the value of a goal opportunity are many, but the most important are the following:

– distance from which the shot is made to the goalpost;

– the angle of the shot;

– the type of shot (by leg, head, knee, heel);

– the rarity of the opportunity, sole opportunity with the goal-keeper, in which the quarterback is not impeded in any way by the defender;

– type of goal pass (diagonal, deep, meet etc.);

– fixed stages;

– dribbles (defender, several players, goal-keeper etc.).

The xG model is especially interesting for sports analysts and commentators, but it is yet not clear how much value it brings to clubs. First of all, these need customized information, at player level, not combined.For example, Stoke shall not take into account the fact that xG for a cross is under 0.1 and shall repeatedly center to Crouch, as will Coutinho aim for the goalpost from outside the square at 16 meters, knowing he has a better chance at success than the implicit xG. Even so, the big English clubs, already have departments specialized on statistics and game predictions.

If in the past we analyzed the goal opportunities and the dangers of shot to the goalpost, now we are already speaking about aggregating these indicators. As to better understand the way in which the xG model works, the chart is extremely explicit for las season:

The best indicators Goals xG xG difference
On players with over 50 shot to the goalpost are included in the top
Kane (Tottenham) 29 18.59 10.41
Lukaku (Everton) 25 15.32 9.68
Llorente (Swansea) 15 7.09 7.91
Son (Tottenham) 14 6.73 7.27
King (Bournemouth) 16 9.56 6.44


The comparisons for the Premier League 2016-2017 season:

Blue for the scored goals, green for xG, expected goals. It is a natural difference, given the fact that the model analyzed the matches played in the past. The biggest difference is for Harry Kane, while Aguero fell into the predictions.

It is a remarkable model, constantly improved by analysts. Even though it does not offer predictions, it gives a general picture of the team and of the way it manages its scoring opportunities. It is considered to be a refined indicator, however, for Europe’s first 10 teams, when goal opportunities, team evolution and achieved points are set in the balanced and analyzed.

Leave a Reply

Your email address will not be published. Required fields are marked *