enter image description here
Expected goals are a great thing. Now let's explain.

It was 2023. Many still need to figure out what xG means (it is also an expected goals). And this is not only about the fans and commentators but also managers.

Everything is relatively easy.

xG is a metric that reflects chances created based on shots on goal. That is, each hit is given a score. For example, Rodri's goal against Bayern Munich from 27 meters is only 0.04 xG (which means that the probability of a goal at the moment of impact is 4%). And Benzema'sBenzema's ball from a meter into an empty net is 0.97 xG (the probability of a goal is 97%). Yes, it is believed that three out of a hundred people would have struck above (to the left/right) of the gate.
enter image description here
enter image description here
Usually, at this moment, in the comments, they start to write: "Are you aware of the matches when one team created more xG, and the other team won?" Yes, you know. This sometimes happens, but here lies the central misunderstanding of the essence. There is no perfect metric in the world that reflects a team's performance in a particular game.

The funny thing is that no one is embarrassed by conversations like: "These struck 10 times, and these - only one." But if someone hints at the expected goals, it often begins: "They don't prove anything!" Such statements sound ridiculous because xG is the same shots; they were classified from 1 to 100. This cannot and should not explain the content of the entire match because there are many subtle elements of the game in football. The fact that the attacker took the defender behind him and freed up the zone for the partner is not calculated and cannot be compared with the benefits of other actions.

As a rule, the team that creates more expected goals wins. However, this does not exclude the situation when City scores an order of magnitude more in terms of xG but ends up in a draw with Everton (1:1 in December 2022). This is not a situation where you need to poke at xG and grin that the metric does not work. She is pretty revealing. If you haven't even watched the broadcast, then looking at the expected goals (1.7 vs. 0.1), you roughly understand how lucky the Liverpool team was in the game.

And now about the minuses of xG - they certainly exist. First, the statistics xG and xPoints (expected points) are indicative precisely at a long distance. It is worth taking statistics for the season, or at least for the lap, to notice trends in players and teams. For example, in the 19th round of the Premier League, Haaland created 0.9 xG but never scored against Chelsea. Although if we take the statistics after 29 matches, we will understand that the Norwegian is just a conversion monster (30 goals in the Premier League at 21.7 xG).
enter image description here
The major bug of expected goals is chances without strikes. Imagine a low-flank shot. The ball rolls along the gate, and the striker barely reaches out to score in the empty ones. A great chance, but xG - 0, because there was no hit. Serious disadvantage. The only good news is that, as a rule, profound moments still end with a blow. Well, again, here is the question of distance. In one match, such cases can distort the picture of the game, but if we take the statistics for a round, these moments will be close to the error level.

And one more drawback. Different statistical companies calculate xG differently. Everyone considers the distance to the goal, the number of defenders on target, and the part of the body that is being kicked, but some companies also make allowances for who exactly shoots and who plays in the goal. And also, for example, FBref estimates a penalty at 0.79 xG and Understat at 0.76 xG. There is a difference in the calculations, but, as a rule, it is insignificant even within the exact match. Yes, and other company indicators may differ - the number of strokes, martial arts, and strikes.

It is exciting whether the xG indicator will help professional bettors to increase their chances of winning. A question that still needs to be answered.